2015-01-05 19:30:17 +00:00
/// @file
/// @author rfree (current maintainer/user in monero.cc project - most of code is from CryptoNote)
/// @brief This is the orginal cryptonote protocol network-events handler, modified by us
2015-01-02 16:52:46 +00:00
// Copyright (c) 2014-2015, The Monero Project
2014-07-23 13:03:52 +00:00
//
// All rights reserved.
//
// Redistribution and use in source and binary forms, with or without modification, are
// permitted provided that the following conditions are met:
//
// 1. Redistributions of source code must retain the above copyright notice, this list of
// conditions and the following disclaimer.
//
// 2. Redistributions in binary form must reproduce the above copyright notice, this list
// of conditions and the following disclaimer in the documentation and/or other
// materials provided with the distribution.
//
// 3. Neither the name of the copyright holder nor the names of its contributors may be
// used to endorse or promote products derived from this software without specific
// prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY
// EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
// MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL
// THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
// INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
// STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF
// THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
// Parts of this file are originally copyright (c) 2012-2013 The Cryptonote developers
2014-03-03 22:07:58 +00:00
2015-01-05 19:30:17 +00:00
// (may contain code and/or modifications by other developers)
// developer rfree: this code is caller of our new network code, and is modded; e.g. for rate limiting
2014-03-03 22:07:58 +00:00
#include <boost/interprocess/detail/atomic.hpp>
2014-07-18 23:33:03 +00:00
#include <list>
2014-03-03 22:07:58 +00:00
#include "cryptonote_core/cryptonote_format_utils.h"
#include "profile_tools.h"
2015-01-05 19:30:17 +00:00
#include "../../contrib/otshell_utils/utils.hpp"
2015-02-12 19:59:39 +00:00
#include "../../src/p2p/network_throttle-detail.hpp"
2015-02-20 21:28:03 +00:00
#include "../../src/p2p/data_logger.hpp"
2015-01-05 19:30:17 +00:00
using namespace nOT::nUtils;
2014-03-03 22:07:58 +00:00
namespace cryptonote
{
2015-01-05 19:30:17 +00:00
// static
// template<class t_core> std::ofstream t_cryptonote_protocol_handler<t_core>::m_logreq("logreq.txt"); // static
2014-03-03 22:07:58 +00:00
//-----------------------------------------------------------------------------------------------------------------------
template<class t_core>
t_cryptonote_protocol_handler<t_core>::t_cryptonote_protocol_handler(t_core& rcore, nodetool::i_p2p_endpoint<connection_context>* p_net_layout):m_core(rcore),
m_p2p(p_net_layout),
m_syncronized_connections_count(0),
2014-03-20 11:46:11 +00:00
m_synchronized(false)
2014-03-03 22:07:58 +00:00
{
if(!m_p2p)
m_p2p = &m_p2p_stub;
}
//-----------------------------------------------------------------------------------------------------------------------
template<class t_core>
bool t_cryptonote_protocol_handler<t_core>::init(const boost::program_options::variables_map& vm)
{
return true;
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
bool t_cryptonote_protocol_handler<t_core>::deinit()
{
return true;
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
void t_cryptonote_protocol_handler<t_core>::set_p2p_endpoint(nodetool::i_p2p_endpoint<connection_context>* p2p)
{
if(p2p)
m_p2p = p2p;
else
m_p2p = &m_p2p_stub;
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
bool t_cryptonote_protocol_handler<t_core>::on_callback(cryptonote_connection_context& context)
{
LOG_PRINT_CCONTEXT_L2("callback fired");
CHECK_AND_ASSERT_MES_CC( context.m_callback_request_count > 0, false, "false callback fired, but context.m_callback_request_count=" << context.m_callback_request_count);
--context.m_callback_request_count;
if(context.m_state == cryptonote_connection_context::state_synchronizing)
{
NOTIFY_REQUEST_CHAIN::request r = boost::value_initialized<NOTIFY_REQUEST_CHAIN::request>();
m_core.get_short_chain_history(r.block_ids);
LOG_PRINT_CCONTEXT_L2("-->>NOTIFY_REQUEST_CHAIN: m_block_ids.size()=" << r.block_ids.size() );
post_notify<NOTIFY_REQUEST_CHAIN>(r, context);
}
return true;
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
bool t_cryptonote_protocol_handler<t_core>::get_stat_info(core_stat_info& stat_inf)
{
return m_core.get_stat_info(stat_inf);
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
2014-03-20 11:46:11 +00:00
void t_cryptonote_protocol_handler<t_core>::log_connections()
{
std::stringstream ss;
2015-02-12 19:59:39 +00:00
ss.precision(1);
double down_sum = 0.0;
double down_curr_sum = 0.0;
double up_sum = 0.0;
double up_curr_sum = 0.0;
2014-03-20 11:46:11 +00:00
2015-01-05 19:30:17 +00:00
ss << std::setw(30) << std::left << "Remote Host"
2014-03-20 11:46:11 +00:00
<< std::setw(20) << "Peer id"
2015-02-12 19:59:39 +00:00
<< std::setw(30) << "Recv/Sent (inactive,sec)"
2014-03-20 11:46:11 +00:00
<< std::setw(25) << "State"
2015-02-12 19:59:39 +00:00
<< std::setw(20) << "Livetime(sec)"
<< std::setw(12) << "Down (kB/s)"
<< std::setw(14) << "Down(now)"
<< std::setw(10) << "Up (kB/s)"
<< std::setw(13) << "Up(now)"
<< ENDL;
2014-03-20 11:46:11 +00:00
2015-01-05 19:30:17 +00:00
uint32_t ip;
2014-03-20 11:46:11 +00:00
m_p2p->for_each_connection([&](const connection_context& cntxt, nodetool::peerid_type peer_id)
{
2015-02-12 19:59:39 +00:00
bool local_ip = false;
2015-01-05 19:30:17 +00:00
ip = ntohl(cntxt.m_remote_ip);
2015-02-12 19:59:39 +00:00
// TODO: local ip in calss A, B
if (ip > 3232235520 && ip < 3232301055) // 192.168.x.x
local_ip = true;
auto connection_time = time(NULL) - cntxt.m_started;
2015-01-05 19:30:17 +00:00
ss << std::setw(30) << std::left << std::string(cntxt.m_is_income ? " [INC]":"[OUT]") +
epee::string_tools::get_ip_string_from_int32(cntxt.m_remote_ip) + ":" + std::to_string(cntxt.m_remote_port)
2014-03-20 11:46:11 +00:00
<< std::setw(20) << std::hex << peer_id
2015-02-12 19:59:39 +00:00
<< std::setw(30) << std::to_string(cntxt.m_recv_cnt)+ "(" + std::to_string(time(NULL) - cntxt.m_last_recv) + ")" + "/" + std::to_string(cntxt.m_send_cnt) + "(" + std::to_string(time(NULL) - cntxt.m_last_send) + ")"
2014-03-20 11:46:11 +00:00
<< std::setw(25) << get_protocol_state_string(cntxt.m_state)
2015-01-05 19:30:17 +00:00
<< std::setw(20) << std::to_string(time(NULL) - cntxt.m_started)
2015-02-12 19:59:39 +00:00
<< std::setw(12) << std::fixed << (connection_time == 0 ? 0.0 : cntxt.m_recv_cnt / connection_time / 1024)
<< std::setw(14) << std::fixed << cntxt.m_current_speed_down / 1024
<< std::setw(10) << std::fixed << (connection_time == 0 ? 0.0 : cntxt.m_send_cnt / connection_time / 1024)
<< std::setw(13) << std::fixed << cntxt.m_current_speed_up / 1024
<< (local_ip ? "[LAN]" : "")
2015-02-20 21:28:03 +00:00
<< std::left << (ip == LOCALHOST_INT ? "[LOCALHOST]" : "") // 127.0.0.1
2015-02-12 19:59:39 +00:00
<< ENDL;
2015-02-20 21:28:03 +00:00
if (connection_time > 1)
2015-02-12 19:59:39 +00:00
{
down_sum += (cntxt.m_recv_cnt / connection_time / 1024);
up_sum += (cntxt.m_send_cnt / connection_time / 1024);
}
down_curr_sum += (cntxt.m_current_speed_down / 1024);
up_curr_sum += (cntxt.m_current_speed_up / 1024);
2014-03-20 11:46:11 +00:00
return true;
});
2015-02-12 19:59:39 +00:00
ss << ENDL
<< std::setw(125) << " "
<< std::setw(12) << down_sum
<< std::setw(14) << down_curr_sum
<< std::setw(10) << up_sum
<< std::setw(13) << up_curr_sum
<< ENDL;
LOG_PRINT_L0("Connections: " << ENDL << ss.str());
2014-03-20 11:46:11 +00:00
}
2014-07-18 23:33:03 +00:00
//------------------------------------------------------------------------------------------------------------------------
// Returns a list of connection_info objects describing each open p2p connection
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
std::list<connection_info> t_cryptonote_protocol_handler<t_core>::get_connections()
{
std::list<connection_info> connections;
m_p2p->for_each_connection([&](const connection_context& cntxt, nodetool::peerid_type peer_id)
{
connection_info cnx;
auto timestamp = time(NULL);
cnx.incoming = cntxt.m_is_income ? true : false;
cnx.ip = epee::string_tools::get_ip_string_from_int32(cntxt.m_remote_ip);
cnx.port = std::to_string(cntxt.m_remote_port);
std::stringstream peer_id_str;
peer_id_str << std::hex << peer_id;
peer_id_str >> cnx.peer_id;
cnx.recv_count = cntxt.m_recv_cnt;
cnx.recv_idle_time = timestamp - cntxt.m_last_recv;
cnx.send_count = cntxt.m_send_cnt;
cnx.send_idle_time = timestamp;
cnx.state = get_protocol_state_string(cntxt.m_state);
cnx.live_time = timestamp - cntxt.m_started;
2015-04-01 17:00:45 +00:00
uint32_t ip;
ip = ntohl(cntxt.m_remote_ip);
if (ip == LOCALHOST_INT)
{
cnx.localhost = true;
}
else
{
cnx.localhost = false;
}
if (ip > 3232235520 && ip < 3232301055) // 192.168.x.x
{
cnx.local_ip = true;
}
else
{
cnx.local_ip = false;
}
auto connection_time = time(NULL) - cntxt.m_started;
if (connection_time == 0)
{
cnx.avg_download = 0;
cnx.avg_upload = 0;
}
else
{
cnx.avg_download = cntxt.m_recv_cnt / connection_time / 1024;
cnx.avg_upload = cntxt.m_send_cnt / connection_time / 1024;
}
cnx.current_download = cntxt.m_current_speed_down / 1024;
cnx.current_upload = cntxt.m_current_speed_up / 1024;
2014-07-18 23:33:03 +00:00
connections.push_back(cnx);
return true;
});
return connections;
}
2014-03-20 11:46:11 +00:00
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
2014-03-03 22:07:58 +00:00
bool t_cryptonote_protocol_handler<t_core>::process_payload_sync_data(const CORE_SYNC_DATA& hshd, cryptonote_connection_context& context, bool is_inital)
{
if(context.m_state == cryptonote_connection_context::state_befor_handshake && !is_inital)
return true;
if(context.m_state == cryptonote_connection_context::state_synchronizing)
return true;
if(m_core.have_block(hshd.top_id))
{
context.m_state = cryptonote_connection_context::state_normal;
if(is_inital)
on_connection_synchronized();
return true;
}
2014-06-04 20:50:13 +00:00
/* As I don't know if accessing hshd from core could be a good practice,
I prefer pushing target height to the core at the same time it is pushed to the user.
Nz. */
2014-06-04 21:10:33 +00:00
m_core.set_target_blockchain_height(static_cast<int64_t>(hshd.current_height));
2014-06-04 20:50:13 +00:00
2014-05-25 17:06:40 +00:00
int64_t diff = static_cast<int64_t>(hshd.current_height) - static_cast<int64_t>(m_core.get_current_blockchain_height());
LOG_PRINT_CCONTEXT_YELLOW("Sync data returned unknown top block: " << m_core.get_current_blockchain_height() << " -> " << hshd.current_height
<< " [" << std::abs(diff) << " blocks (" << diff / (24 * 60 * 60 / DIFFICULTY_TARGET) << " days) "
<< (0 <= diff ? std::string("behind") : std::string("ahead"))
<< "] " << ENDL << "SYNCHRONIZATION started", (is_inital ? LOG_LEVEL_0:LOG_LEVEL_1));
LOG_PRINT_L1("Remote top block height: " << hshd.current_height << ", id: " << hshd.top_id);
2014-03-03 22:07:58 +00:00
context.m_state = cryptonote_connection_context::state_synchronizing;
context.m_remote_blockchain_height = hshd.current_height;
//let the socket to send response to handshake, but request callback, to let send request data after response
LOG_PRINT_CCONTEXT_L2("requesting callback");
++context.m_callback_request_count;
m_p2p->request_callback(context);
return true;
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
bool t_cryptonote_protocol_handler<t_core>::get_payload_sync_data(CORE_SYNC_DATA& hshd)
{
m_core.get_blockchain_top(hshd.current_height, hshd.top_id);
hshd.current_height +=1;
return true;
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
bool t_cryptonote_protocol_handler<t_core>::get_payload_sync_data(blobdata& data)
{
CORE_SYNC_DATA hsd = boost::value_initialized<CORE_SYNC_DATA>();
get_payload_sync_data(hsd);
epee::serialization::store_t_to_binary(hsd, data);
return true;
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
int t_cryptonote_protocol_handler<t_core>::handle_notify_new_block(int command, NOTIFY_NEW_BLOCK::request& arg, cryptonote_connection_context& context)
{
LOG_PRINT_CCONTEXT_L2("NOTIFY_NEW_BLOCK (hop " << arg.hop << ")");
if(context.m_state != cryptonote_connection_context::state_normal)
return 1;
** CHANGES ARE EXPERIMENTAL (FOR TESTING ONLY)
Bockchain:
1. Optim: Multi-thread long-hash computation when encountering groups of blocks.
2. Optim: Cache verified txs and return result from cache instead of re-checking whenever possible.
3. Optim: Preload output-keys when encoutering groups of blocks. Sort by amount and global-index before bulk querying database and multi-thread when possible.
4. Optim: Disable double spend check on block verification, double spend is already detected when trying to add blocks.
5. Optim: Multi-thread signature computation whenever possible.
6. Patch: Disable locking (recursive mutex) on called functions from check_tx_inputs which causes slowdowns (only seems to happen on ubuntu/VMs??? Reason: TBD)
7. Optim: Removed looped full-tx hash computation when retrieving transactions from pool (???).
8. Optim: Cache difficulty/timestamps (735 blocks) for next-difficulty calculations so that only 2 db reads per new block is needed when a new block arrives (instead of 1470 reads).
Berkeley-DB:
1. Fix: 32-bit data errors causing wrong output global indices and failure to send blocks to peers (etc).
2. Fix: Unable to pop blocks on reorganize due to transaction errors.
3. Patch: Large number of transaction aborts when running multi-threaded bulk queries.
4. Patch: Insufficient locks error when running full sync.
5. Patch: Incorrect db stats when returning from an immediate exit from "pop block" operation.
6. Optim: Add bulk queries to get output global indices.
7. Optim: Modified output_keys table to store public_key+unlock_time+height for single transaction lookup (vs 3)
8. Optim: Used output_keys table retrieve public_keys instead of going through output_amounts->output_txs+output_indices->txs->output:public_key
9. Optim: Added thread-safe buffers used when multi-threading bulk queries.
10. Optim: Added support for nosync/write_nosync options for improved performance (*see --db-sync-mode option for details)
11. Mod: Added checkpoint thread and auto-remove-logs option.
12. *Now usable on 32-bit systems like RPI2.
LMDB:
1. Optim: Added custom comparison for 256-bit key tables (minor speed-up, TBD: get actual effect)
2. Optim: Modified output_keys table to store public_key+unlock_time+height for single transaction lookup (vs 3)
3. Optim: Used output_keys table retrieve public_keys instead of going through output_amounts->output_txs+output_indices->txs->output:public_key
4. Optim: Added support for sync/writemap options for improved performance (*see --db-sync-mode option for details)
5. Mod: Auto resize to +1GB instead of multiplier x1.5
ETC:
1. Minor optimizations for slow-hash for ARM (RPI2). Incomplete.
2. Fix: 32-bit saturation bug when computing next difficulty on large blocks.
[PENDING ISSUES]
1. Berkely db has a very slow "pop-block" operation. This is very noticeable on the RPI2 as it sometimes takes > 10 MINUTES to pop a block during reorganization.
This does not happen very often however, most reorgs seem to take a few seconds but it possibly depends on the number of outputs present. TBD.
2. Berkeley db, possible bug "unable to allocate memory". TBD.
[NEW OPTIONS] (*Currently all enabled for testing purposes)
1. --fast-block-sync arg=[0:1] (default: 1)
a. 0 = Compute long hash per block (may take a while depending on CPU)
b. 1 = Skip long-hash and verify blocks based on embedded known good block hashes (faster, minimal CPU dependence)
2. --db-sync-mode arg=[[safe|fast|fastest]:[sync|async]:[nblocks_per_sync]] (default: fastest:async:1000)
a. safe = fdatasync/fsync (or equivalent) per stored block. Very slow, but safest option to protect against power-out/crash conditions.
b. fast/fastest = Enables asynchronous fdatasync/fsync (or equivalent). Useful for battery operated devices or STABLE systems with UPS and/or systems with battery backed write cache/solid state cache.
Fast - Write meta-data but defer data flush.
Fastest - Defer meta-data and data flush.
Sync - Flush data after nblocks_per_sync and wait.
Async - Flush data after nblocks_per_sync but do not wait for the operation to finish.
3. --prep-blocks-threads arg=[n] (default: 4 or system max threads, whichever is lower)
Max number of threads to use when computing long-hash in groups.
4. --show-time-stats arg=[0:1] (default: 1)
Show benchmark related time stats.
5. --db-auto-remove-logs arg=[0:1] (default: 1)
For berkeley-db only. Auto remove logs if enabled.
**Note: lmdb and berkeley-db have changes to the tables and are not compatible with official git head version.
At the moment, you need a full resync to use this optimized version.
[PERFORMANCE COMPARISON]
**Some figures are approximations only.
Using a baseline machine of an i7-2600K+SSD+(with full pow computation):
1. The optimized lmdb/blockhain core can process blocks up to 585K for ~1.25 hours + download time, so it usually takes 2.5 hours to sync the full chain.
2. The current head with memory can process blocks up to 585K for ~4.2 hours + download time, so it usually takes 5.5 hours to sync the full chain.
3. The current head with lmdb can process blocks up to 585K for ~32 hours + download time and usually takes 36 hours to sync the full chain.
Averate procesing times (with full pow computation):
lmdb-optimized:
1. tx_ave = 2.5 ms / tx
2. block_ave = 5.87 ms / block
memory-official-repo:
1. tx_ave = 8.85 ms / tx
2. block_ave = 19.68 ms / block
lmdb-official-repo (0f4a036437fd41a5498ee5e74e2422ea6177aa3e)
1. tx_ave = 47.8 ms / tx
2. block_ave = 64.2 ms / block
**Note: The following data denotes processing times only (does not include p2p download time)
lmdb-optimized processing times (with full pow computation):
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 1.25 hours processing time (--db-sync-mode=fastest:async:1000).
2. Laptop, Dual-core / 4-threads U4200 (3Mb) - 4.90 hours processing time (--db-sync-mode=fastest:async:1000).
3. Embedded, Quad-core / 4-threads Z3735F (2x1Mb) - 12.0 hours processing time (--db-sync-mode=fastest:async:1000).
lmdb-optimized processing times (with per-block-checkpoint)
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 10 minutes processing time (--db-sync-mode=fastest:async:1000).
berkeley-db optimized processing times (with full pow computation)
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 1.8 hours processing time (--db-sync-mode=fastest:async:1000).
2. RPI2. Improved from estimated 3 months(???) into 2.5 days (*Need 2AMP supply + Clock:1Ghz + [usb+ssd] to achieve this speed) (--db-sync-mode=fastest:async:1000).
berkeley-db optimized processing times (with per-block-checkpoint)
1. RPI2. 12-15 hours (*Need 2AMP supply + Clock:1Ghz + [usb+ssd] to achieve this speed) (--db-sync-mode=fastest:async:1000).
2015-07-10 20:09:32 +00:00
std::list<block_complete_entry> blocks;
blocks.push_back(arg.b);
m_core.prepare_handle_incoming_blocks(blocks);
2014-03-03 22:07:58 +00:00
for(auto tx_blob_it = arg.b.txs.begin(); tx_blob_it!=arg.b.txs.end();tx_blob_it++)
{
cryptonote::tx_verification_context tvc = AUTO_VAL_INIT(tvc);
m_core.handle_incoming_tx(*tx_blob_it, tvc, true);
if(tvc.m_verifivation_failed)
{
2014-09-09 09:32:00 +00:00
LOG_PRINT_CCONTEXT_L1("Block verification failed: transaction verification failed, dropping connection");
2014-03-03 22:07:58 +00:00
m_p2p->drop_connection(context);
** CHANGES ARE EXPERIMENTAL (FOR TESTING ONLY)
Bockchain:
1. Optim: Multi-thread long-hash computation when encountering groups of blocks.
2. Optim: Cache verified txs and return result from cache instead of re-checking whenever possible.
3. Optim: Preload output-keys when encoutering groups of blocks. Sort by amount and global-index before bulk querying database and multi-thread when possible.
4. Optim: Disable double spend check on block verification, double spend is already detected when trying to add blocks.
5. Optim: Multi-thread signature computation whenever possible.
6. Patch: Disable locking (recursive mutex) on called functions from check_tx_inputs which causes slowdowns (only seems to happen on ubuntu/VMs??? Reason: TBD)
7. Optim: Removed looped full-tx hash computation when retrieving transactions from pool (???).
8. Optim: Cache difficulty/timestamps (735 blocks) for next-difficulty calculations so that only 2 db reads per new block is needed when a new block arrives (instead of 1470 reads).
Berkeley-DB:
1. Fix: 32-bit data errors causing wrong output global indices and failure to send blocks to peers (etc).
2. Fix: Unable to pop blocks on reorganize due to transaction errors.
3. Patch: Large number of transaction aborts when running multi-threaded bulk queries.
4. Patch: Insufficient locks error when running full sync.
5. Patch: Incorrect db stats when returning from an immediate exit from "pop block" operation.
6. Optim: Add bulk queries to get output global indices.
7. Optim: Modified output_keys table to store public_key+unlock_time+height for single transaction lookup (vs 3)
8. Optim: Used output_keys table retrieve public_keys instead of going through output_amounts->output_txs+output_indices->txs->output:public_key
9. Optim: Added thread-safe buffers used when multi-threading bulk queries.
10. Optim: Added support for nosync/write_nosync options for improved performance (*see --db-sync-mode option for details)
11. Mod: Added checkpoint thread and auto-remove-logs option.
12. *Now usable on 32-bit systems like RPI2.
LMDB:
1. Optim: Added custom comparison for 256-bit key tables (minor speed-up, TBD: get actual effect)
2. Optim: Modified output_keys table to store public_key+unlock_time+height for single transaction lookup (vs 3)
3. Optim: Used output_keys table retrieve public_keys instead of going through output_amounts->output_txs+output_indices->txs->output:public_key
4. Optim: Added support for sync/writemap options for improved performance (*see --db-sync-mode option for details)
5. Mod: Auto resize to +1GB instead of multiplier x1.5
ETC:
1. Minor optimizations for slow-hash for ARM (RPI2). Incomplete.
2. Fix: 32-bit saturation bug when computing next difficulty on large blocks.
[PENDING ISSUES]
1. Berkely db has a very slow "pop-block" operation. This is very noticeable on the RPI2 as it sometimes takes > 10 MINUTES to pop a block during reorganization.
This does not happen very often however, most reorgs seem to take a few seconds but it possibly depends on the number of outputs present. TBD.
2. Berkeley db, possible bug "unable to allocate memory". TBD.
[NEW OPTIONS] (*Currently all enabled for testing purposes)
1. --fast-block-sync arg=[0:1] (default: 1)
a. 0 = Compute long hash per block (may take a while depending on CPU)
b. 1 = Skip long-hash and verify blocks based on embedded known good block hashes (faster, minimal CPU dependence)
2. --db-sync-mode arg=[[safe|fast|fastest]:[sync|async]:[nblocks_per_sync]] (default: fastest:async:1000)
a. safe = fdatasync/fsync (or equivalent) per stored block. Very slow, but safest option to protect against power-out/crash conditions.
b. fast/fastest = Enables asynchronous fdatasync/fsync (or equivalent). Useful for battery operated devices or STABLE systems with UPS and/or systems with battery backed write cache/solid state cache.
Fast - Write meta-data but defer data flush.
Fastest - Defer meta-data and data flush.
Sync - Flush data after nblocks_per_sync and wait.
Async - Flush data after nblocks_per_sync but do not wait for the operation to finish.
3. --prep-blocks-threads arg=[n] (default: 4 or system max threads, whichever is lower)
Max number of threads to use when computing long-hash in groups.
4. --show-time-stats arg=[0:1] (default: 1)
Show benchmark related time stats.
5. --db-auto-remove-logs arg=[0:1] (default: 1)
For berkeley-db only. Auto remove logs if enabled.
**Note: lmdb and berkeley-db have changes to the tables and are not compatible with official git head version.
At the moment, you need a full resync to use this optimized version.
[PERFORMANCE COMPARISON]
**Some figures are approximations only.
Using a baseline machine of an i7-2600K+SSD+(with full pow computation):
1. The optimized lmdb/blockhain core can process blocks up to 585K for ~1.25 hours + download time, so it usually takes 2.5 hours to sync the full chain.
2. The current head with memory can process blocks up to 585K for ~4.2 hours + download time, so it usually takes 5.5 hours to sync the full chain.
3. The current head with lmdb can process blocks up to 585K for ~32 hours + download time and usually takes 36 hours to sync the full chain.
Averate procesing times (with full pow computation):
lmdb-optimized:
1. tx_ave = 2.5 ms / tx
2. block_ave = 5.87 ms / block
memory-official-repo:
1. tx_ave = 8.85 ms / tx
2. block_ave = 19.68 ms / block
lmdb-official-repo (0f4a036437fd41a5498ee5e74e2422ea6177aa3e)
1. tx_ave = 47.8 ms / tx
2. block_ave = 64.2 ms / block
**Note: The following data denotes processing times only (does not include p2p download time)
lmdb-optimized processing times (with full pow computation):
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 1.25 hours processing time (--db-sync-mode=fastest:async:1000).
2. Laptop, Dual-core / 4-threads U4200 (3Mb) - 4.90 hours processing time (--db-sync-mode=fastest:async:1000).
3. Embedded, Quad-core / 4-threads Z3735F (2x1Mb) - 12.0 hours processing time (--db-sync-mode=fastest:async:1000).
lmdb-optimized processing times (with per-block-checkpoint)
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 10 minutes processing time (--db-sync-mode=fastest:async:1000).
berkeley-db optimized processing times (with full pow computation)
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 1.8 hours processing time (--db-sync-mode=fastest:async:1000).
2. RPI2. Improved from estimated 3 months(???) into 2.5 days (*Need 2AMP supply + Clock:1Ghz + [usb+ssd] to achieve this speed) (--db-sync-mode=fastest:async:1000).
berkeley-db optimized processing times (with per-block-checkpoint)
1. RPI2. 12-15 hours (*Need 2AMP supply + Clock:1Ghz + [usb+ssd] to achieve this speed) (--db-sync-mode=fastest:async:1000).
2015-07-10 20:09:32 +00:00
m_core.cleanup_handle_incoming_blocks();
2014-03-03 22:07:58 +00:00
return 1;
}
}
block_verification_context bvc = boost::value_initialized<block_verification_context>();
m_core.pause_mine();
2015-01-05 19:30:17 +00:00
m_core.handle_incoming_block(arg.b.block, bvc); // got block from handle_notify_new_block
** CHANGES ARE EXPERIMENTAL (FOR TESTING ONLY)
Bockchain:
1. Optim: Multi-thread long-hash computation when encountering groups of blocks.
2. Optim: Cache verified txs and return result from cache instead of re-checking whenever possible.
3. Optim: Preload output-keys when encoutering groups of blocks. Sort by amount and global-index before bulk querying database and multi-thread when possible.
4. Optim: Disable double spend check on block verification, double spend is already detected when trying to add blocks.
5. Optim: Multi-thread signature computation whenever possible.
6. Patch: Disable locking (recursive mutex) on called functions from check_tx_inputs which causes slowdowns (only seems to happen on ubuntu/VMs??? Reason: TBD)
7. Optim: Removed looped full-tx hash computation when retrieving transactions from pool (???).
8. Optim: Cache difficulty/timestamps (735 blocks) for next-difficulty calculations so that only 2 db reads per new block is needed when a new block arrives (instead of 1470 reads).
Berkeley-DB:
1. Fix: 32-bit data errors causing wrong output global indices and failure to send blocks to peers (etc).
2. Fix: Unable to pop blocks on reorganize due to transaction errors.
3. Patch: Large number of transaction aborts when running multi-threaded bulk queries.
4. Patch: Insufficient locks error when running full sync.
5. Patch: Incorrect db stats when returning from an immediate exit from "pop block" operation.
6. Optim: Add bulk queries to get output global indices.
7. Optim: Modified output_keys table to store public_key+unlock_time+height for single transaction lookup (vs 3)
8. Optim: Used output_keys table retrieve public_keys instead of going through output_amounts->output_txs+output_indices->txs->output:public_key
9. Optim: Added thread-safe buffers used when multi-threading bulk queries.
10. Optim: Added support for nosync/write_nosync options for improved performance (*see --db-sync-mode option for details)
11. Mod: Added checkpoint thread and auto-remove-logs option.
12. *Now usable on 32-bit systems like RPI2.
LMDB:
1. Optim: Added custom comparison for 256-bit key tables (minor speed-up, TBD: get actual effect)
2. Optim: Modified output_keys table to store public_key+unlock_time+height for single transaction lookup (vs 3)
3. Optim: Used output_keys table retrieve public_keys instead of going through output_amounts->output_txs+output_indices->txs->output:public_key
4. Optim: Added support for sync/writemap options for improved performance (*see --db-sync-mode option for details)
5. Mod: Auto resize to +1GB instead of multiplier x1.5
ETC:
1. Minor optimizations for slow-hash for ARM (RPI2). Incomplete.
2. Fix: 32-bit saturation bug when computing next difficulty on large blocks.
[PENDING ISSUES]
1. Berkely db has a very slow "pop-block" operation. This is very noticeable on the RPI2 as it sometimes takes > 10 MINUTES to pop a block during reorganization.
This does not happen very often however, most reorgs seem to take a few seconds but it possibly depends on the number of outputs present. TBD.
2. Berkeley db, possible bug "unable to allocate memory". TBD.
[NEW OPTIONS] (*Currently all enabled for testing purposes)
1. --fast-block-sync arg=[0:1] (default: 1)
a. 0 = Compute long hash per block (may take a while depending on CPU)
b. 1 = Skip long-hash and verify blocks based on embedded known good block hashes (faster, minimal CPU dependence)
2. --db-sync-mode arg=[[safe|fast|fastest]:[sync|async]:[nblocks_per_sync]] (default: fastest:async:1000)
a. safe = fdatasync/fsync (or equivalent) per stored block. Very slow, but safest option to protect against power-out/crash conditions.
b. fast/fastest = Enables asynchronous fdatasync/fsync (or equivalent). Useful for battery operated devices or STABLE systems with UPS and/or systems with battery backed write cache/solid state cache.
Fast - Write meta-data but defer data flush.
Fastest - Defer meta-data and data flush.
Sync - Flush data after nblocks_per_sync and wait.
Async - Flush data after nblocks_per_sync but do not wait for the operation to finish.
3. --prep-blocks-threads arg=[n] (default: 4 or system max threads, whichever is lower)
Max number of threads to use when computing long-hash in groups.
4. --show-time-stats arg=[0:1] (default: 1)
Show benchmark related time stats.
5. --db-auto-remove-logs arg=[0:1] (default: 1)
For berkeley-db only. Auto remove logs if enabled.
**Note: lmdb and berkeley-db have changes to the tables and are not compatible with official git head version.
At the moment, you need a full resync to use this optimized version.
[PERFORMANCE COMPARISON]
**Some figures are approximations only.
Using a baseline machine of an i7-2600K+SSD+(with full pow computation):
1. The optimized lmdb/blockhain core can process blocks up to 585K for ~1.25 hours + download time, so it usually takes 2.5 hours to sync the full chain.
2. The current head with memory can process blocks up to 585K for ~4.2 hours + download time, so it usually takes 5.5 hours to sync the full chain.
3. The current head with lmdb can process blocks up to 585K for ~32 hours + download time and usually takes 36 hours to sync the full chain.
Averate procesing times (with full pow computation):
lmdb-optimized:
1. tx_ave = 2.5 ms / tx
2. block_ave = 5.87 ms / block
memory-official-repo:
1. tx_ave = 8.85 ms / tx
2. block_ave = 19.68 ms / block
lmdb-official-repo (0f4a036437fd41a5498ee5e74e2422ea6177aa3e)
1. tx_ave = 47.8 ms / tx
2. block_ave = 64.2 ms / block
**Note: The following data denotes processing times only (does not include p2p download time)
lmdb-optimized processing times (with full pow computation):
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 1.25 hours processing time (--db-sync-mode=fastest:async:1000).
2. Laptop, Dual-core / 4-threads U4200 (3Mb) - 4.90 hours processing time (--db-sync-mode=fastest:async:1000).
3. Embedded, Quad-core / 4-threads Z3735F (2x1Mb) - 12.0 hours processing time (--db-sync-mode=fastest:async:1000).
lmdb-optimized processing times (with per-block-checkpoint)
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 10 minutes processing time (--db-sync-mode=fastest:async:1000).
berkeley-db optimized processing times (with full pow computation)
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 1.8 hours processing time (--db-sync-mode=fastest:async:1000).
2. RPI2. Improved from estimated 3 months(???) into 2.5 days (*Need 2AMP supply + Clock:1Ghz + [usb+ssd] to achieve this speed) (--db-sync-mode=fastest:async:1000).
berkeley-db optimized processing times (with per-block-checkpoint)
1. RPI2. 12-15 hours (*Need 2AMP supply + Clock:1Ghz + [usb+ssd] to achieve this speed) (--db-sync-mode=fastest:async:1000).
2015-07-10 20:09:32 +00:00
m_core.cleanup_handle_incoming_blocks(true);
2014-03-03 22:07:58 +00:00
m_core.resume_mine();
if(bvc.m_verifivation_failed)
{
2015-01-05 19:30:17 +00:00
LOG_PRINT_CCONTEXT_L0("Block verification failed, dropping connection");
2014-03-03 22:07:58 +00:00
m_p2p->drop_connection(context);
return 1;
}
if(bvc.m_added_to_main_chain)
{
++arg.hop;
//TODO: Add here announce protocol usage
relay_block(arg, context);
}else if(bvc.m_marked_as_orphaned)
{
context.m_state = cryptonote_connection_context::state_synchronizing;
NOTIFY_REQUEST_CHAIN::request r = boost::value_initialized<NOTIFY_REQUEST_CHAIN::request>();
m_core.get_short_chain_history(r.block_ids);
LOG_PRINT_CCONTEXT_L2("-->>NOTIFY_REQUEST_CHAIN: m_block_ids.size()=" << r.block_ids.size() );
post_notify<NOTIFY_REQUEST_CHAIN>(r, context);
}
return 1;
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
int t_cryptonote_protocol_handler<t_core>::handle_notify_new_transactions(int command, NOTIFY_NEW_TRANSACTIONS::request& arg, cryptonote_connection_context& context)
{
LOG_PRINT_CCONTEXT_L2("NOTIFY_NEW_TRANSACTIONS");
if(context.m_state != cryptonote_connection_context::state_normal)
return 1;
for(auto tx_blob_it = arg.txs.begin(); tx_blob_it!=arg.txs.end();)
{
cryptonote::tx_verification_context tvc = AUTO_VAL_INIT(tvc);
m_core.handle_incoming_tx(*tx_blob_it, tvc, false);
if(tvc.m_verifivation_failed)
{
2014-09-09 09:32:00 +00:00
LOG_PRINT_CCONTEXT_L1("Tx verification failed, dropping connection");
2014-03-03 22:07:58 +00:00
m_p2p->drop_connection(context);
return 1;
}
if(tvc.m_should_be_relayed)
++tx_blob_it;
else
arg.txs.erase(tx_blob_it++);
}
if(arg.txs.size())
{
//TODO: add announce usage here
relay_transactions(arg, context);
}
return true;
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
int t_cryptonote_protocol_handler<t_core>::handle_request_get_objects(int command, NOTIFY_REQUEST_GET_OBJECTS::request& arg, cryptonote_connection_context& context)
{
LOG_PRINT_CCONTEXT_L2("NOTIFY_REQUEST_GET_OBJECTS");
NOTIFY_RESPONSE_GET_OBJECTS::request rsp;
if(!m_core.handle_get_objects(arg, rsp, context))
{
LOG_ERROR_CCONTEXT("failed to handle request NOTIFY_REQUEST_GET_OBJECTS, dropping connection");
m_p2p->drop_connection(context);
}
LOG_PRINT_CCONTEXT_L2("-->>NOTIFY_RESPONSE_GET_OBJECTS: blocks.size()=" << rsp.blocks.size() << ", txs.size()=" << rsp.txs.size()
<< ", rsp.m_current_blockchain_height=" << rsp.current_blockchain_height << ", missed_ids.size()=" << rsp.missed_ids.size());
post_notify<NOTIFY_RESPONSE_GET_OBJECTS>(rsp, context);
2015-01-05 19:30:17 +00:00
//handler_response_blocks_now(sizeof(rsp)); // XXX
//handler_response_blocks_now(200);
2014-03-03 22:07:58 +00:00
return 1;
}
//------------------------------------------------------------------------------------------------------------------------
2015-01-05 19:30:17 +00:00
template<class t_core>
2015-02-12 19:59:39 +00:00
double t_cryptonote_protocol_handler<t_core>::get_avg_block_size() {
// return m_core.get_blockchain_storage().get_avg_block_size(count); // this does not count too well the actuall network-size of data we need to download
CRITICAL_REGION_LOCAL(m_buffer_mutex);
double avg = 0;
if (m_avg_buffer.size() == 0) {
_warn("m_avg_buffer.size() == 0");
return 500;
}
const bool dbg_poke_lock = 0; // debug: try to trigger an error by poking around with locks. TODO: configure option
long int dbg_repeat=0;
do {
for (auto element : m_avg_buffer) avg += element;
} while(dbg_poke_lock && (dbg_repeat++)<100000); // in debug/poke mode, repeat this calculation to trigger hidden locking error if there is one
return avg / m_avg_buffer.size();
2015-01-05 19:30:17 +00:00
}
2014-03-03 22:07:58 +00:00
template<class t_core>
int t_cryptonote_protocol_handler<t_core>::handle_response_get_objects(int command, NOTIFY_RESPONSE_GET_OBJECTS::request& arg, cryptonote_connection_context& context)
{
LOG_PRINT_CCONTEXT_L2("NOTIFY_RESPONSE_GET_OBJECTS");
2015-02-12 19:59:39 +00:00
// calculate size of request - mainly for logging/debug
size_t size = 0;
2015-02-20 21:28:03 +00:00
for (auto element : arg.txs) size += element.size();
2015-02-12 19:59:39 +00:00
2015-02-20 21:28:03 +00:00
for (auto element : arg.blocks) {
2015-02-12 19:59:39 +00:00
size += element.block.size();
for (auto tx : element.txs)
size += tx.size();
}
for (auto element : arg.missed_ids)
size += sizeof(element.data);
size += sizeof(arg.current_blockchain_height);
{
CRITICAL_REGION_LOCAL(m_buffer_mutex);
m_avg_buffer.push_back(size);
const bool dbg_poke_lock = 0; // debug: try to trigger an error by poking around with locks. TODO: configure option
long int dbg_repeat=0;
do {
m_avg_buffer.push_back(666); // a test value
m_avg_buffer.erase_end(1);
} while(dbg_poke_lock && (dbg_repeat++)<100000); // in debug/poke mode, repeat this calculation to trigger hidden locking error if there is one
}
/*using namespace boost::chrono;
auto point = steady_clock::now();
auto time_from_epoh = point.time_since_epoch();
auto sec = duration_cast< seconds >( time_from_epoh ).count();*/
//epee::net_utils::network_throttle_manager::get_global_throttle_inreq().logger_handle_net("log/dr-monero/net/req-all.data", sec, get_avg_block_size());
2014-03-03 22:07:58 +00:00
if(context.m_last_response_height > arg.current_blockchain_height)
{
LOG_ERROR_CCONTEXT("sent wrong NOTIFY_HAVE_OBJECTS: arg.m_current_blockchain_height=" << arg.current_blockchain_height
<< " < m_last_response_height=" << context.m_last_response_height << ", dropping connection");
m_p2p->drop_connection(context);
return 1;
}
context.m_remote_blockchain_height = arg.current_blockchain_height;
2014-03-20 11:46:11 +00:00
size_t count = 0;
2014-03-03 22:07:58 +00:00
BOOST_FOREACH(const block_complete_entry& block_entry, arg.blocks)
{
2014-03-20 11:46:11 +00:00
++count;
2014-03-03 22:07:58 +00:00
block b;
if(!parse_and_validate_block_from_blob(block_entry.block, b))
{
LOG_ERROR_CCONTEXT("sent wrong block: failed to parse and validate block: \r\n"
2014-05-25 17:06:40 +00:00
<< epee::string_tools::buff_to_hex_nodelimer(block_entry.block) << "\r\n dropping connection");
2014-03-03 22:07:58 +00:00
m_p2p->drop_connection(context);
return 1;
2014-03-20 11:46:11 +00:00
}
//to avoid concurrency in core between connections, suspend connections which delivered block later then first one
if(count == 2)
{
if(m_core.have_block(get_block_hash(b)))
{
context.m_state = cryptonote_connection_context::state_idle;
context.m_needed_objects.clear();
context.m_requested_objects.clear();
LOG_PRINT_CCONTEXT_L1("Connection set to idle state.");
return 1;
}
2014-03-03 22:07:58 +00:00
}
auto req_it = context.m_requested_objects.find(get_block_hash(b));
if(req_it == context.m_requested_objects.end())
{
2014-05-25 17:06:40 +00:00
LOG_ERROR_CCONTEXT("sent wrong NOTIFY_RESPONSE_GET_OBJECTS: block with id=" << epee::string_tools::pod_to_hex(get_blob_hash(block_entry.block))
2014-03-03 22:07:58 +00:00
<< " wasn't requested, dropping connection");
m_p2p->drop_connection(context);
return 1;
}
if(b.tx_hashes.size() != block_entry.txs.size())
{
2014-05-25 17:06:40 +00:00
LOG_ERROR_CCONTEXT("sent wrong NOTIFY_RESPONSE_GET_OBJECTS: block with id=" << epee::string_tools::pod_to_hex(get_blob_hash(block_entry.block))
2014-03-03 22:07:58 +00:00
<< ", tx_hashes.size()=" << b.tx_hashes.size() << " mismatch with block_complete_entry.m_txs.size()=" << block_entry.txs.size() << ", dropping connection");
m_p2p->drop_connection(context);
return 1;
}
context.m_requested_objects.erase(req_it);
}
if(context.m_requested_objects.size())
{
LOG_PRINT_CCONTEXT_RED("returned not all requested objects (context.m_requested_objects.size()="
<< context.m_requested_objects.size() << "), dropping connection", LOG_LEVEL_0);
m_p2p->drop_connection(context);
return 1;
}
2015-02-20 21:28:03 +00:00
2014-03-03 22:07:58 +00:00
{
m_core.pause_mine();
2014-05-25 17:06:40 +00:00
epee::misc_utils::auto_scope_leave_caller scope_exit_handler = epee::misc_utils::create_scope_leave_handler(
2014-03-03 22:07:58 +00:00
boost::bind(&t_core::resume_mine, &m_core));
2015-01-05 19:30:17 +00:00
LOG_PRINT_CCONTEXT_YELLOW( "Got NEW BLOCKS inside of " << __FUNCTION__ << ": size: " << arg.blocks.size() , LOG_LEVEL_0);
2015-02-20 21:28:03 +00:00
if (m_core.get_test_drop_download() && m_core.get_test_drop_download_height()) { // DISCARD BLOCKS for testing
** CHANGES ARE EXPERIMENTAL (FOR TESTING ONLY)
Bockchain:
1. Optim: Multi-thread long-hash computation when encountering groups of blocks.
2. Optim: Cache verified txs and return result from cache instead of re-checking whenever possible.
3. Optim: Preload output-keys when encoutering groups of blocks. Sort by amount and global-index before bulk querying database and multi-thread when possible.
4. Optim: Disable double spend check on block verification, double spend is already detected when trying to add blocks.
5. Optim: Multi-thread signature computation whenever possible.
6. Patch: Disable locking (recursive mutex) on called functions from check_tx_inputs which causes slowdowns (only seems to happen on ubuntu/VMs??? Reason: TBD)
7. Optim: Removed looped full-tx hash computation when retrieving transactions from pool (???).
8. Optim: Cache difficulty/timestamps (735 blocks) for next-difficulty calculations so that only 2 db reads per new block is needed when a new block arrives (instead of 1470 reads).
Berkeley-DB:
1. Fix: 32-bit data errors causing wrong output global indices and failure to send blocks to peers (etc).
2. Fix: Unable to pop blocks on reorganize due to transaction errors.
3. Patch: Large number of transaction aborts when running multi-threaded bulk queries.
4. Patch: Insufficient locks error when running full sync.
5. Patch: Incorrect db stats when returning from an immediate exit from "pop block" operation.
6. Optim: Add bulk queries to get output global indices.
7. Optim: Modified output_keys table to store public_key+unlock_time+height for single transaction lookup (vs 3)
8. Optim: Used output_keys table retrieve public_keys instead of going through output_amounts->output_txs+output_indices->txs->output:public_key
9. Optim: Added thread-safe buffers used when multi-threading bulk queries.
10. Optim: Added support for nosync/write_nosync options for improved performance (*see --db-sync-mode option for details)
11. Mod: Added checkpoint thread and auto-remove-logs option.
12. *Now usable on 32-bit systems like RPI2.
LMDB:
1. Optim: Added custom comparison for 256-bit key tables (minor speed-up, TBD: get actual effect)
2. Optim: Modified output_keys table to store public_key+unlock_time+height for single transaction lookup (vs 3)
3. Optim: Used output_keys table retrieve public_keys instead of going through output_amounts->output_txs+output_indices->txs->output:public_key
4. Optim: Added support for sync/writemap options for improved performance (*see --db-sync-mode option for details)
5. Mod: Auto resize to +1GB instead of multiplier x1.5
ETC:
1. Minor optimizations for slow-hash for ARM (RPI2). Incomplete.
2. Fix: 32-bit saturation bug when computing next difficulty on large blocks.
[PENDING ISSUES]
1. Berkely db has a very slow "pop-block" operation. This is very noticeable on the RPI2 as it sometimes takes > 10 MINUTES to pop a block during reorganization.
This does not happen very often however, most reorgs seem to take a few seconds but it possibly depends on the number of outputs present. TBD.
2. Berkeley db, possible bug "unable to allocate memory". TBD.
[NEW OPTIONS] (*Currently all enabled for testing purposes)
1. --fast-block-sync arg=[0:1] (default: 1)
a. 0 = Compute long hash per block (may take a while depending on CPU)
b. 1 = Skip long-hash and verify blocks based on embedded known good block hashes (faster, minimal CPU dependence)
2. --db-sync-mode arg=[[safe|fast|fastest]:[sync|async]:[nblocks_per_sync]] (default: fastest:async:1000)
a. safe = fdatasync/fsync (or equivalent) per stored block. Very slow, but safest option to protect against power-out/crash conditions.
b. fast/fastest = Enables asynchronous fdatasync/fsync (or equivalent). Useful for battery operated devices or STABLE systems with UPS and/or systems with battery backed write cache/solid state cache.
Fast - Write meta-data but defer data flush.
Fastest - Defer meta-data and data flush.
Sync - Flush data after nblocks_per_sync and wait.
Async - Flush data after nblocks_per_sync but do not wait for the operation to finish.
3. --prep-blocks-threads arg=[n] (default: 4 or system max threads, whichever is lower)
Max number of threads to use when computing long-hash in groups.
4. --show-time-stats arg=[0:1] (default: 1)
Show benchmark related time stats.
5. --db-auto-remove-logs arg=[0:1] (default: 1)
For berkeley-db only. Auto remove logs if enabled.
**Note: lmdb and berkeley-db have changes to the tables and are not compatible with official git head version.
At the moment, you need a full resync to use this optimized version.
[PERFORMANCE COMPARISON]
**Some figures are approximations only.
Using a baseline machine of an i7-2600K+SSD+(with full pow computation):
1. The optimized lmdb/blockhain core can process blocks up to 585K for ~1.25 hours + download time, so it usually takes 2.5 hours to sync the full chain.
2. The current head with memory can process blocks up to 585K for ~4.2 hours + download time, so it usually takes 5.5 hours to sync the full chain.
3. The current head with lmdb can process blocks up to 585K for ~32 hours + download time and usually takes 36 hours to sync the full chain.
Averate procesing times (with full pow computation):
lmdb-optimized:
1. tx_ave = 2.5 ms / tx
2. block_ave = 5.87 ms / block
memory-official-repo:
1. tx_ave = 8.85 ms / tx
2. block_ave = 19.68 ms / block
lmdb-official-repo (0f4a036437fd41a5498ee5e74e2422ea6177aa3e)
1. tx_ave = 47.8 ms / tx
2. block_ave = 64.2 ms / block
**Note: The following data denotes processing times only (does not include p2p download time)
lmdb-optimized processing times (with full pow computation):
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 1.25 hours processing time (--db-sync-mode=fastest:async:1000).
2. Laptop, Dual-core / 4-threads U4200 (3Mb) - 4.90 hours processing time (--db-sync-mode=fastest:async:1000).
3. Embedded, Quad-core / 4-threads Z3735F (2x1Mb) - 12.0 hours processing time (--db-sync-mode=fastest:async:1000).
lmdb-optimized processing times (with per-block-checkpoint)
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 10 minutes processing time (--db-sync-mode=fastest:async:1000).
berkeley-db optimized processing times (with full pow computation)
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 1.8 hours processing time (--db-sync-mode=fastest:async:1000).
2. RPI2. Improved from estimated 3 months(???) into 2.5 days (*Need 2AMP supply + Clock:1Ghz + [usb+ssd] to achieve this speed) (--db-sync-mode=fastest:async:1000).
berkeley-db optimized processing times (with per-block-checkpoint)
1. RPI2. 12-15 hours (*Need 2AMP supply + Clock:1Ghz + [usb+ssd] to achieve this speed) (--db-sync-mode=fastest:async:1000).
2015-07-10 20:09:32 +00:00
m_core.prepare_handle_incoming_blocks(arg.blocks);
2015-02-20 21:28:03 +00:00
BOOST_FOREACH(const block_complete_entry& block_entry, arg.blocks)
{
// process transactions
TIME_MEASURE_START(transactions_process_time);
BOOST_FOREACH(auto& tx_blob, block_entry.txs)
{
tx_verification_context tvc = AUTO_VAL_INIT(tvc);
m_core.handle_incoming_tx(tx_blob, tvc, true);
if(tvc.m_verifivation_failed)
{
LOG_ERROR_CCONTEXT("transaction verification failed on NOTIFY_RESPONSE_GET_OBJECTS, \r\ntx_id = "
<< epee::string_tools::pod_to_hex(get_blob_hash(tx_blob)) << ", dropping connection");
m_p2p->drop_connection(context);
** CHANGES ARE EXPERIMENTAL (FOR TESTING ONLY)
Bockchain:
1. Optim: Multi-thread long-hash computation when encountering groups of blocks.
2. Optim: Cache verified txs and return result from cache instead of re-checking whenever possible.
3. Optim: Preload output-keys when encoutering groups of blocks. Sort by amount and global-index before bulk querying database and multi-thread when possible.
4. Optim: Disable double spend check on block verification, double spend is already detected when trying to add blocks.
5. Optim: Multi-thread signature computation whenever possible.
6. Patch: Disable locking (recursive mutex) on called functions from check_tx_inputs which causes slowdowns (only seems to happen on ubuntu/VMs??? Reason: TBD)
7. Optim: Removed looped full-tx hash computation when retrieving transactions from pool (???).
8. Optim: Cache difficulty/timestamps (735 blocks) for next-difficulty calculations so that only 2 db reads per new block is needed when a new block arrives (instead of 1470 reads).
Berkeley-DB:
1. Fix: 32-bit data errors causing wrong output global indices and failure to send blocks to peers (etc).
2. Fix: Unable to pop blocks on reorganize due to transaction errors.
3. Patch: Large number of transaction aborts when running multi-threaded bulk queries.
4. Patch: Insufficient locks error when running full sync.
5. Patch: Incorrect db stats when returning from an immediate exit from "pop block" operation.
6. Optim: Add bulk queries to get output global indices.
7. Optim: Modified output_keys table to store public_key+unlock_time+height for single transaction lookup (vs 3)
8. Optim: Used output_keys table retrieve public_keys instead of going through output_amounts->output_txs+output_indices->txs->output:public_key
9. Optim: Added thread-safe buffers used when multi-threading bulk queries.
10. Optim: Added support for nosync/write_nosync options for improved performance (*see --db-sync-mode option for details)
11. Mod: Added checkpoint thread and auto-remove-logs option.
12. *Now usable on 32-bit systems like RPI2.
LMDB:
1. Optim: Added custom comparison for 256-bit key tables (minor speed-up, TBD: get actual effect)
2. Optim: Modified output_keys table to store public_key+unlock_time+height for single transaction lookup (vs 3)
3. Optim: Used output_keys table retrieve public_keys instead of going through output_amounts->output_txs+output_indices->txs->output:public_key
4. Optim: Added support for sync/writemap options for improved performance (*see --db-sync-mode option for details)
5. Mod: Auto resize to +1GB instead of multiplier x1.5
ETC:
1. Minor optimizations for slow-hash for ARM (RPI2). Incomplete.
2. Fix: 32-bit saturation bug when computing next difficulty on large blocks.
[PENDING ISSUES]
1. Berkely db has a very slow "pop-block" operation. This is very noticeable on the RPI2 as it sometimes takes > 10 MINUTES to pop a block during reorganization.
This does not happen very often however, most reorgs seem to take a few seconds but it possibly depends on the number of outputs present. TBD.
2. Berkeley db, possible bug "unable to allocate memory". TBD.
[NEW OPTIONS] (*Currently all enabled for testing purposes)
1. --fast-block-sync arg=[0:1] (default: 1)
a. 0 = Compute long hash per block (may take a while depending on CPU)
b. 1 = Skip long-hash and verify blocks based on embedded known good block hashes (faster, minimal CPU dependence)
2. --db-sync-mode arg=[[safe|fast|fastest]:[sync|async]:[nblocks_per_sync]] (default: fastest:async:1000)
a. safe = fdatasync/fsync (or equivalent) per stored block. Very slow, but safest option to protect against power-out/crash conditions.
b. fast/fastest = Enables asynchronous fdatasync/fsync (or equivalent). Useful for battery operated devices or STABLE systems with UPS and/or systems with battery backed write cache/solid state cache.
Fast - Write meta-data but defer data flush.
Fastest - Defer meta-data and data flush.
Sync - Flush data after nblocks_per_sync and wait.
Async - Flush data after nblocks_per_sync but do not wait for the operation to finish.
3. --prep-blocks-threads arg=[n] (default: 4 or system max threads, whichever is lower)
Max number of threads to use when computing long-hash in groups.
4. --show-time-stats arg=[0:1] (default: 1)
Show benchmark related time stats.
5. --db-auto-remove-logs arg=[0:1] (default: 1)
For berkeley-db only. Auto remove logs if enabled.
**Note: lmdb and berkeley-db have changes to the tables and are not compatible with official git head version.
At the moment, you need a full resync to use this optimized version.
[PERFORMANCE COMPARISON]
**Some figures are approximations only.
Using a baseline machine of an i7-2600K+SSD+(with full pow computation):
1. The optimized lmdb/blockhain core can process blocks up to 585K for ~1.25 hours + download time, so it usually takes 2.5 hours to sync the full chain.
2. The current head with memory can process blocks up to 585K for ~4.2 hours + download time, so it usually takes 5.5 hours to sync the full chain.
3. The current head with lmdb can process blocks up to 585K for ~32 hours + download time and usually takes 36 hours to sync the full chain.
Averate procesing times (with full pow computation):
lmdb-optimized:
1. tx_ave = 2.5 ms / tx
2. block_ave = 5.87 ms / block
memory-official-repo:
1. tx_ave = 8.85 ms / tx
2. block_ave = 19.68 ms / block
lmdb-official-repo (0f4a036437fd41a5498ee5e74e2422ea6177aa3e)
1. tx_ave = 47.8 ms / tx
2. block_ave = 64.2 ms / block
**Note: The following data denotes processing times only (does not include p2p download time)
lmdb-optimized processing times (with full pow computation):
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 1.25 hours processing time (--db-sync-mode=fastest:async:1000).
2. Laptop, Dual-core / 4-threads U4200 (3Mb) - 4.90 hours processing time (--db-sync-mode=fastest:async:1000).
3. Embedded, Quad-core / 4-threads Z3735F (2x1Mb) - 12.0 hours processing time (--db-sync-mode=fastest:async:1000).
lmdb-optimized processing times (with per-block-checkpoint)
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 10 minutes processing time (--db-sync-mode=fastest:async:1000).
berkeley-db optimized processing times (with full pow computation)
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 1.8 hours processing time (--db-sync-mode=fastest:async:1000).
2. RPI2. Improved from estimated 3 months(???) into 2.5 days (*Need 2AMP supply + Clock:1Ghz + [usb+ssd] to achieve this speed) (--db-sync-mode=fastest:async:1000).
berkeley-db optimized processing times (with per-block-checkpoint)
1. RPI2. 12-15 hours (*Need 2AMP supply + Clock:1Ghz + [usb+ssd] to achieve this speed) (--db-sync-mode=fastest:async:1000).
2015-07-10 20:09:32 +00:00
m_core.cleanup_handle_incoming_blocks();
2015-02-20 21:28:03 +00:00
return 1;
}
}
TIME_MEASURE_FINISH(transactions_process_time);
// process block
TIME_MEASURE_START(block_process_time);
block_verification_context bvc = boost::value_initialized<block_verification_context>();
m_core.handle_incoming_block(block_entry.block, bvc, false); // <--- process block
if(bvc.m_verifivation_failed)
{
LOG_PRINT_CCONTEXT_L1("Block verification failed, dropping connection");
m_p2p->drop_connection(context);
** CHANGES ARE EXPERIMENTAL (FOR TESTING ONLY)
Bockchain:
1. Optim: Multi-thread long-hash computation when encountering groups of blocks.
2. Optim: Cache verified txs and return result from cache instead of re-checking whenever possible.
3. Optim: Preload output-keys when encoutering groups of blocks. Sort by amount and global-index before bulk querying database and multi-thread when possible.
4. Optim: Disable double spend check on block verification, double spend is already detected when trying to add blocks.
5. Optim: Multi-thread signature computation whenever possible.
6. Patch: Disable locking (recursive mutex) on called functions from check_tx_inputs which causes slowdowns (only seems to happen on ubuntu/VMs??? Reason: TBD)
7. Optim: Removed looped full-tx hash computation when retrieving transactions from pool (???).
8. Optim: Cache difficulty/timestamps (735 blocks) for next-difficulty calculations so that only 2 db reads per new block is needed when a new block arrives (instead of 1470 reads).
Berkeley-DB:
1. Fix: 32-bit data errors causing wrong output global indices and failure to send blocks to peers (etc).
2. Fix: Unable to pop blocks on reorganize due to transaction errors.
3. Patch: Large number of transaction aborts when running multi-threaded bulk queries.
4. Patch: Insufficient locks error when running full sync.
5. Patch: Incorrect db stats when returning from an immediate exit from "pop block" operation.
6. Optim: Add bulk queries to get output global indices.
7. Optim: Modified output_keys table to store public_key+unlock_time+height for single transaction lookup (vs 3)
8. Optim: Used output_keys table retrieve public_keys instead of going through output_amounts->output_txs+output_indices->txs->output:public_key
9. Optim: Added thread-safe buffers used when multi-threading bulk queries.
10. Optim: Added support for nosync/write_nosync options for improved performance (*see --db-sync-mode option for details)
11. Mod: Added checkpoint thread and auto-remove-logs option.
12. *Now usable on 32-bit systems like RPI2.
LMDB:
1. Optim: Added custom comparison for 256-bit key tables (minor speed-up, TBD: get actual effect)
2. Optim: Modified output_keys table to store public_key+unlock_time+height for single transaction lookup (vs 3)
3. Optim: Used output_keys table retrieve public_keys instead of going through output_amounts->output_txs+output_indices->txs->output:public_key
4. Optim: Added support for sync/writemap options for improved performance (*see --db-sync-mode option for details)
5. Mod: Auto resize to +1GB instead of multiplier x1.5
ETC:
1. Minor optimizations for slow-hash for ARM (RPI2). Incomplete.
2. Fix: 32-bit saturation bug when computing next difficulty on large blocks.
[PENDING ISSUES]
1. Berkely db has a very slow "pop-block" operation. This is very noticeable on the RPI2 as it sometimes takes > 10 MINUTES to pop a block during reorganization.
This does not happen very often however, most reorgs seem to take a few seconds but it possibly depends on the number of outputs present. TBD.
2. Berkeley db, possible bug "unable to allocate memory". TBD.
[NEW OPTIONS] (*Currently all enabled for testing purposes)
1. --fast-block-sync arg=[0:1] (default: 1)
a. 0 = Compute long hash per block (may take a while depending on CPU)
b. 1 = Skip long-hash and verify blocks based on embedded known good block hashes (faster, minimal CPU dependence)
2. --db-sync-mode arg=[[safe|fast|fastest]:[sync|async]:[nblocks_per_sync]] (default: fastest:async:1000)
a. safe = fdatasync/fsync (or equivalent) per stored block. Very slow, but safest option to protect against power-out/crash conditions.
b. fast/fastest = Enables asynchronous fdatasync/fsync (or equivalent). Useful for battery operated devices or STABLE systems with UPS and/or systems with battery backed write cache/solid state cache.
Fast - Write meta-data but defer data flush.
Fastest - Defer meta-data and data flush.
Sync - Flush data after nblocks_per_sync and wait.
Async - Flush data after nblocks_per_sync but do not wait for the operation to finish.
3. --prep-blocks-threads arg=[n] (default: 4 or system max threads, whichever is lower)
Max number of threads to use when computing long-hash in groups.
4. --show-time-stats arg=[0:1] (default: 1)
Show benchmark related time stats.
5. --db-auto-remove-logs arg=[0:1] (default: 1)
For berkeley-db only. Auto remove logs if enabled.
**Note: lmdb and berkeley-db have changes to the tables and are not compatible with official git head version.
At the moment, you need a full resync to use this optimized version.
[PERFORMANCE COMPARISON]
**Some figures are approximations only.
Using a baseline machine of an i7-2600K+SSD+(with full pow computation):
1. The optimized lmdb/blockhain core can process blocks up to 585K for ~1.25 hours + download time, so it usually takes 2.5 hours to sync the full chain.
2. The current head with memory can process blocks up to 585K for ~4.2 hours + download time, so it usually takes 5.5 hours to sync the full chain.
3. The current head with lmdb can process blocks up to 585K for ~32 hours + download time and usually takes 36 hours to sync the full chain.
Averate procesing times (with full pow computation):
lmdb-optimized:
1. tx_ave = 2.5 ms / tx
2. block_ave = 5.87 ms / block
memory-official-repo:
1. tx_ave = 8.85 ms / tx
2. block_ave = 19.68 ms / block
lmdb-official-repo (0f4a036437fd41a5498ee5e74e2422ea6177aa3e)
1. tx_ave = 47.8 ms / tx
2. block_ave = 64.2 ms / block
**Note: The following data denotes processing times only (does not include p2p download time)
lmdb-optimized processing times (with full pow computation):
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 1.25 hours processing time (--db-sync-mode=fastest:async:1000).
2. Laptop, Dual-core / 4-threads U4200 (3Mb) - 4.90 hours processing time (--db-sync-mode=fastest:async:1000).
3. Embedded, Quad-core / 4-threads Z3735F (2x1Mb) - 12.0 hours processing time (--db-sync-mode=fastest:async:1000).
lmdb-optimized processing times (with per-block-checkpoint)
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 10 minutes processing time (--db-sync-mode=fastest:async:1000).
berkeley-db optimized processing times (with full pow computation)
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 1.8 hours processing time (--db-sync-mode=fastest:async:1000).
2. RPI2. Improved from estimated 3 months(???) into 2.5 days (*Need 2AMP supply + Clock:1Ghz + [usb+ssd] to achieve this speed) (--db-sync-mode=fastest:async:1000).
berkeley-db optimized processing times (with per-block-checkpoint)
1. RPI2. 12-15 hours (*Need 2AMP supply + Clock:1Ghz + [usb+ssd] to achieve this speed) (--db-sync-mode=fastest:async:1000).
2015-07-10 20:09:32 +00:00
m_core.cleanup_handle_incoming_blocks();
2015-02-20 21:28:03 +00:00
return 1;
}
if(bvc.m_marked_as_orphaned)
{
LOG_PRINT_CCONTEXT_L1("Block received at sync phase was marked as orphaned, dropping connection");
m_p2p->drop_connection(context);
** CHANGES ARE EXPERIMENTAL (FOR TESTING ONLY)
Bockchain:
1. Optim: Multi-thread long-hash computation when encountering groups of blocks.
2. Optim: Cache verified txs and return result from cache instead of re-checking whenever possible.
3. Optim: Preload output-keys when encoutering groups of blocks. Sort by amount and global-index before bulk querying database and multi-thread when possible.
4. Optim: Disable double spend check on block verification, double spend is already detected when trying to add blocks.
5. Optim: Multi-thread signature computation whenever possible.
6. Patch: Disable locking (recursive mutex) on called functions from check_tx_inputs which causes slowdowns (only seems to happen on ubuntu/VMs??? Reason: TBD)
7. Optim: Removed looped full-tx hash computation when retrieving transactions from pool (???).
8. Optim: Cache difficulty/timestamps (735 blocks) for next-difficulty calculations so that only 2 db reads per new block is needed when a new block arrives (instead of 1470 reads).
Berkeley-DB:
1. Fix: 32-bit data errors causing wrong output global indices and failure to send blocks to peers (etc).
2. Fix: Unable to pop blocks on reorganize due to transaction errors.
3. Patch: Large number of transaction aborts when running multi-threaded bulk queries.
4. Patch: Insufficient locks error when running full sync.
5. Patch: Incorrect db stats when returning from an immediate exit from "pop block" operation.
6. Optim: Add bulk queries to get output global indices.
7. Optim: Modified output_keys table to store public_key+unlock_time+height for single transaction lookup (vs 3)
8. Optim: Used output_keys table retrieve public_keys instead of going through output_amounts->output_txs+output_indices->txs->output:public_key
9. Optim: Added thread-safe buffers used when multi-threading bulk queries.
10. Optim: Added support for nosync/write_nosync options for improved performance (*see --db-sync-mode option for details)
11. Mod: Added checkpoint thread and auto-remove-logs option.
12. *Now usable on 32-bit systems like RPI2.
LMDB:
1. Optim: Added custom comparison for 256-bit key tables (minor speed-up, TBD: get actual effect)
2. Optim: Modified output_keys table to store public_key+unlock_time+height for single transaction lookup (vs 3)
3. Optim: Used output_keys table retrieve public_keys instead of going through output_amounts->output_txs+output_indices->txs->output:public_key
4. Optim: Added support for sync/writemap options for improved performance (*see --db-sync-mode option for details)
5. Mod: Auto resize to +1GB instead of multiplier x1.5
ETC:
1. Minor optimizations for slow-hash for ARM (RPI2). Incomplete.
2. Fix: 32-bit saturation bug when computing next difficulty on large blocks.
[PENDING ISSUES]
1. Berkely db has a very slow "pop-block" operation. This is very noticeable on the RPI2 as it sometimes takes > 10 MINUTES to pop a block during reorganization.
This does not happen very often however, most reorgs seem to take a few seconds but it possibly depends on the number of outputs present. TBD.
2. Berkeley db, possible bug "unable to allocate memory". TBD.
[NEW OPTIONS] (*Currently all enabled for testing purposes)
1. --fast-block-sync arg=[0:1] (default: 1)
a. 0 = Compute long hash per block (may take a while depending on CPU)
b. 1 = Skip long-hash and verify blocks based on embedded known good block hashes (faster, minimal CPU dependence)
2. --db-sync-mode arg=[[safe|fast|fastest]:[sync|async]:[nblocks_per_sync]] (default: fastest:async:1000)
a. safe = fdatasync/fsync (or equivalent) per stored block. Very slow, but safest option to protect against power-out/crash conditions.
b. fast/fastest = Enables asynchronous fdatasync/fsync (or equivalent). Useful for battery operated devices or STABLE systems with UPS and/or systems with battery backed write cache/solid state cache.
Fast - Write meta-data but defer data flush.
Fastest - Defer meta-data and data flush.
Sync - Flush data after nblocks_per_sync and wait.
Async - Flush data after nblocks_per_sync but do not wait for the operation to finish.
3. --prep-blocks-threads arg=[n] (default: 4 or system max threads, whichever is lower)
Max number of threads to use when computing long-hash in groups.
4. --show-time-stats arg=[0:1] (default: 1)
Show benchmark related time stats.
5. --db-auto-remove-logs arg=[0:1] (default: 1)
For berkeley-db only. Auto remove logs if enabled.
**Note: lmdb and berkeley-db have changes to the tables and are not compatible with official git head version.
At the moment, you need a full resync to use this optimized version.
[PERFORMANCE COMPARISON]
**Some figures are approximations only.
Using a baseline machine of an i7-2600K+SSD+(with full pow computation):
1. The optimized lmdb/blockhain core can process blocks up to 585K for ~1.25 hours + download time, so it usually takes 2.5 hours to sync the full chain.
2. The current head with memory can process blocks up to 585K for ~4.2 hours + download time, so it usually takes 5.5 hours to sync the full chain.
3. The current head with lmdb can process blocks up to 585K for ~32 hours + download time and usually takes 36 hours to sync the full chain.
Averate procesing times (with full pow computation):
lmdb-optimized:
1. tx_ave = 2.5 ms / tx
2. block_ave = 5.87 ms / block
memory-official-repo:
1. tx_ave = 8.85 ms / tx
2. block_ave = 19.68 ms / block
lmdb-official-repo (0f4a036437fd41a5498ee5e74e2422ea6177aa3e)
1. tx_ave = 47.8 ms / tx
2. block_ave = 64.2 ms / block
**Note: The following data denotes processing times only (does not include p2p download time)
lmdb-optimized processing times (with full pow computation):
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 1.25 hours processing time (--db-sync-mode=fastest:async:1000).
2. Laptop, Dual-core / 4-threads U4200 (3Mb) - 4.90 hours processing time (--db-sync-mode=fastest:async:1000).
3. Embedded, Quad-core / 4-threads Z3735F (2x1Mb) - 12.0 hours processing time (--db-sync-mode=fastest:async:1000).
lmdb-optimized processing times (with per-block-checkpoint)
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 10 minutes processing time (--db-sync-mode=fastest:async:1000).
berkeley-db optimized processing times (with full pow computation)
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 1.8 hours processing time (--db-sync-mode=fastest:async:1000).
2. RPI2. Improved from estimated 3 months(???) into 2.5 days (*Need 2AMP supply + Clock:1Ghz + [usb+ssd] to achieve this speed) (--db-sync-mode=fastest:async:1000).
berkeley-db optimized processing times (with per-block-checkpoint)
1. RPI2. 12-15 hours (*Need 2AMP supply + Clock:1Ghz + [usb+ssd] to achieve this speed) (--db-sync-mode=fastest:async:1000).
2015-07-10 20:09:32 +00:00
m_core.cleanup_handle_incoming_blocks();
2015-02-20 21:28:03 +00:00
return 1;
}
TIME_MEASURE_FINISH(block_process_time);
LOG_PRINT_CCONTEXT_L2("Block process time: " << block_process_time + transactions_process_time << "(" << transactions_process_time << "/" << block_process_time << ")ms");
epee::net_utils::data_logger::get_instance().add_data("calc_time", block_process_time + transactions_process_time);
2015-04-01 17:00:45 +00:00
epee::net_utils::data_logger::get_instance().add_data("block_processing", 1);
2015-02-20 21:28:03 +00:00
} // each download block
** CHANGES ARE EXPERIMENTAL (FOR TESTING ONLY)
Bockchain:
1. Optim: Multi-thread long-hash computation when encountering groups of blocks.
2. Optim: Cache verified txs and return result from cache instead of re-checking whenever possible.
3. Optim: Preload output-keys when encoutering groups of blocks. Sort by amount and global-index before bulk querying database and multi-thread when possible.
4. Optim: Disable double spend check on block verification, double spend is already detected when trying to add blocks.
5. Optim: Multi-thread signature computation whenever possible.
6. Patch: Disable locking (recursive mutex) on called functions from check_tx_inputs which causes slowdowns (only seems to happen on ubuntu/VMs??? Reason: TBD)
7. Optim: Removed looped full-tx hash computation when retrieving transactions from pool (???).
8. Optim: Cache difficulty/timestamps (735 blocks) for next-difficulty calculations so that only 2 db reads per new block is needed when a new block arrives (instead of 1470 reads).
Berkeley-DB:
1. Fix: 32-bit data errors causing wrong output global indices and failure to send blocks to peers (etc).
2. Fix: Unable to pop blocks on reorganize due to transaction errors.
3. Patch: Large number of transaction aborts when running multi-threaded bulk queries.
4. Patch: Insufficient locks error when running full sync.
5. Patch: Incorrect db stats when returning from an immediate exit from "pop block" operation.
6. Optim: Add bulk queries to get output global indices.
7. Optim: Modified output_keys table to store public_key+unlock_time+height for single transaction lookup (vs 3)
8. Optim: Used output_keys table retrieve public_keys instead of going through output_amounts->output_txs+output_indices->txs->output:public_key
9. Optim: Added thread-safe buffers used when multi-threading bulk queries.
10. Optim: Added support for nosync/write_nosync options for improved performance (*see --db-sync-mode option for details)
11. Mod: Added checkpoint thread and auto-remove-logs option.
12. *Now usable on 32-bit systems like RPI2.
LMDB:
1. Optim: Added custom comparison for 256-bit key tables (minor speed-up, TBD: get actual effect)
2. Optim: Modified output_keys table to store public_key+unlock_time+height for single transaction lookup (vs 3)
3. Optim: Used output_keys table retrieve public_keys instead of going through output_amounts->output_txs+output_indices->txs->output:public_key
4. Optim: Added support for sync/writemap options for improved performance (*see --db-sync-mode option for details)
5. Mod: Auto resize to +1GB instead of multiplier x1.5
ETC:
1. Minor optimizations for slow-hash for ARM (RPI2). Incomplete.
2. Fix: 32-bit saturation bug when computing next difficulty on large blocks.
[PENDING ISSUES]
1. Berkely db has a very slow "pop-block" operation. This is very noticeable on the RPI2 as it sometimes takes > 10 MINUTES to pop a block during reorganization.
This does not happen very often however, most reorgs seem to take a few seconds but it possibly depends on the number of outputs present. TBD.
2. Berkeley db, possible bug "unable to allocate memory". TBD.
[NEW OPTIONS] (*Currently all enabled for testing purposes)
1. --fast-block-sync arg=[0:1] (default: 1)
a. 0 = Compute long hash per block (may take a while depending on CPU)
b. 1 = Skip long-hash and verify blocks based on embedded known good block hashes (faster, minimal CPU dependence)
2. --db-sync-mode arg=[[safe|fast|fastest]:[sync|async]:[nblocks_per_sync]] (default: fastest:async:1000)
a. safe = fdatasync/fsync (or equivalent) per stored block. Very slow, but safest option to protect against power-out/crash conditions.
b. fast/fastest = Enables asynchronous fdatasync/fsync (or equivalent). Useful for battery operated devices or STABLE systems with UPS and/or systems with battery backed write cache/solid state cache.
Fast - Write meta-data but defer data flush.
Fastest - Defer meta-data and data flush.
Sync - Flush data after nblocks_per_sync and wait.
Async - Flush data after nblocks_per_sync but do not wait for the operation to finish.
3. --prep-blocks-threads arg=[n] (default: 4 or system max threads, whichever is lower)
Max number of threads to use when computing long-hash in groups.
4. --show-time-stats arg=[0:1] (default: 1)
Show benchmark related time stats.
5. --db-auto-remove-logs arg=[0:1] (default: 1)
For berkeley-db only. Auto remove logs if enabled.
**Note: lmdb and berkeley-db have changes to the tables and are not compatible with official git head version.
At the moment, you need a full resync to use this optimized version.
[PERFORMANCE COMPARISON]
**Some figures are approximations only.
Using a baseline machine of an i7-2600K+SSD+(with full pow computation):
1. The optimized lmdb/blockhain core can process blocks up to 585K for ~1.25 hours + download time, so it usually takes 2.5 hours to sync the full chain.
2. The current head with memory can process blocks up to 585K for ~4.2 hours + download time, so it usually takes 5.5 hours to sync the full chain.
3. The current head with lmdb can process blocks up to 585K for ~32 hours + download time and usually takes 36 hours to sync the full chain.
Averate procesing times (with full pow computation):
lmdb-optimized:
1. tx_ave = 2.5 ms / tx
2. block_ave = 5.87 ms / block
memory-official-repo:
1. tx_ave = 8.85 ms / tx
2. block_ave = 19.68 ms / block
lmdb-official-repo (0f4a036437fd41a5498ee5e74e2422ea6177aa3e)
1. tx_ave = 47.8 ms / tx
2. block_ave = 64.2 ms / block
**Note: The following data denotes processing times only (does not include p2p download time)
lmdb-optimized processing times (with full pow computation):
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 1.25 hours processing time (--db-sync-mode=fastest:async:1000).
2. Laptop, Dual-core / 4-threads U4200 (3Mb) - 4.90 hours processing time (--db-sync-mode=fastest:async:1000).
3. Embedded, Quad-core / 4-threads Z3735F (2x1Mb) - 12.0 hours processing time (--db-sync-mode=fastest:async:1000).
lmdb-optimized processing times (with per-block-checkpoint)
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 10 minutes processing time (--db-sync-mode=fastest:async:1000).
berkeley-db optimized processing times (with full pow computation)
1. Desktop, Quad-core / 8-threads 2600k (8Mb) - 1.8 hours processing time (--db-sync-mode=fastest:async:1000).
2. RPI2. Improved from estimated 3 months(???) into 2.5 days (*Need 2AMP supply + Clock:1Ghz + [usb+ssd] to achieve this speed) (--db-sync-mode=fastest:async:1000).
berkeley-db optimized processing times (with per-block-checkpoint)
1. RPI2. 12-15 hours (*Need 2AMP supply + Clock:1Ghz + [usb+ssd] to achieve this speed) (--db-sync-mode=fastest:async:1000).
2015-07-10 20:09:32 +00:00
m_core.cleanup_handle_incoming_blocks();
2015-02-20 21:28:03 +00:00
} // if not DISCARD BLOCK
2014-03-03 22:07:58 +00:00
}
request_missing_objects(context, true);
return 1;
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
bool t_cryptonote_protocol_handler<t_core>::on_idle()
{
return m_core.on_idle();
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
int t_cryptonote_protocol_handler<t_core>::handle_request_chain(int command, NOTIFY_REQUEST_CHAIN::request& arg, cryptonote_connection_context& context)
{
LOG_PRINT_CCONTEXT_L2("NOTIFY_REQUEST_CHAIN: m_block_ids.size()=" << arg.block_ids.size());
NOTIFY_RESPONSE_CHAIN_ENTRY::request r;
if(!m_core.find_blockchain_supplement(arg.block_ids, r))
{
LOG_ERROR_CCONTEXT("Failed to handle NOTIFY_REQUEST_CHAIN.");
return 1;
}
LOG_PRINT_CCONTEXT_L2("-->>NOTIFY_RESPONSE_CHAIN_ENTRY: m_start_height=" << r.start_height << ", m_total_height=" << r.total_height << ", m_block_ids.size()=" << r.m_block_ids.size());
post_notify<NOTIFY_RESPONSE_CHAIN_ENTRY>(r, context);
return 1;
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
bool t_cryptonote_protocol_handler<t_core>::request_missing_objects(cryptonote_connection_context& context, bool check_having_blocks)
{
2015-02-12 19:59:39 +00:00
//if (!m_one_request == false)
//return true;
m_one_request = false;
// save request size to log (dr monero)
/*using namespace boost::chrono;
auto point = steady_clock::now();
auto time_from_epoh = point.time_since_epoch();
auto sec = duration_cast< seconds >( time_from_epoh ).count();*/
2014-03-03 22:07:58 +00:00
if(context.m_needed_objects.size())
{
//we know objects that we need, request this objects
NOTIFY_REQUEST_GET_OBJECTS::request req;
size_t count = 0;
auto it = context.m_needed_objects.begin();
2015-02-12 19:59:39 +00:00
size_t count_limit = BLOCKS_SYNCHRONIZING_DEFAULT_COUNT;
_note_c("net/req-calc" , "Setting count_limit: " << count_limit);
2014-03-03 22:07:58 +00:00
while(it != context.m_needed_objects.end() && count < BLOCKS_SYNCHRONIZING_DEFAULT_COUNT)
{
if( !(check_having_blocks && m_core.have_block(*it)))
{
req.blocks.push_back(*it);
++count;
context.m_requested_objects.insert(*it);
}
context.m_needed_objects.erase(it++);
}
2015-01-05 19:30:17 +00:00
LOG_PRINT_CCONTEXT_L0("-->>NOTIFY_REQUEST_GET_OBJECTS: blocks.size()=" << req.blocks.size() << ", txs.size()=" << req.txs.size()
<< "requested blocks count=" << count << " / " << count_limit);
2015-02-12 19:59:39 +00:00
//epee::net_utils::network_throttle_manager::get_global_throttle_inreq().logger_handle_net("log/dr-monero/net/req-all.data", sec, get_avg_block_size());
2014-03-03 22:07:58 +00:00
post_notify<NOTIFY_REQUEST_GET_OBJECTS>(req, context);
}else if(context.m_last_response_height < context.m_remote_blockchain_height-1)
{//we have to fetch more objects ids, request blockchain entry
NOTIFY_REQUEST_CHAIN::request r = boost::value_initialized<NOTIFY_REQUEST_CHAIN::request>();
m_core.get_short_chain_history(r.block_ids);
2015-01-05 19:30:17 +00:00
handler_request_blocks_history( r.block_ids ); // change the limit(?), sleep(?)
2015-02-12 19:59:39 +00:00
//std::string blob; // for calculate size of request
//epee::serialization::store_t_to_binary(r, blob);
//epee::net_utils::network_throttle_manager::get_global_throttle_inreq().logger_handle_net("log/dr-monero/net/req-all.data", sec, get_avg_block_size());
LOG_PRINT_CCONTEXT_L0("r = " << 200);
2015-01-05 19:30:17 +00:00
LOG_PRINT_CCONTEXT_L0("-->>NOTIFY_REQUEST_CHAIN: m_block_ids.size()=" << r.block_ids.size() );
2014-03-03 22:07:58 +00:00
post_notify<NOTIFY_REQUEST_CHAIN>(r, context);
}else
{
CHECK_AND_ASSERT_MES(context.m_last_response_height == context.m_remote_blockchain_height-1
&& !context.m_needed_objects.size()
&& !context.m_requested_objects.size(), false, "request_missing_blocks final condition failed!"
<< "\r\nm_last_response_height=" << context.m_last_response_height
<< "\r\nm_remote_blockchain_height=" << context.m_remote_blockchain_height
<< "\r\nm_needed_objects.size()=" << context.m_needed_objects.size()
<< "\r\nm_requested_objects.size()=" << context.m_requested_objects.size()
2014-05-25 17:06:40 +00:00
<< "\r\non connection [" << epee::net_utils::print_connection_context_short(context)<< "]");
2014-03-03 22:07:58 +00:00
context.m_state = cryptonote_connection_context::state_normal;
LOG_PRINT_CCONTEXT_GREEN(" SYNCHRONIZED OK", LOG_LEVEL_0);
2014-03-20 11:46:11 +00:00
on_connection_synchronized();
2014-03-03 22:07:58 +00:00
}
return true;
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
bool t_cryptonote_protocol_handler<t_core>::on_connection_synchronized()
{
bool val_expected = false;
2014-03-20 11:46:11 +00:00
if(m_synchronized.compare_exchange_strong(val_expected, true))
2014-03-03 22:07:58 +00:00
{
LOG_PRINT_L0(ENDL << "**********************************************************************" << ENDL
<< "You are now synchronized with the network. You may now start simplewallet." << ENDL
<< ENDL
<< "Please note, that the blockchain will be saved only after you quit the daemon with \"exit\" command or if you use \"save\" command." << ENDL
<< "Otherwise, you will possibly need to synchronize the blockchain again." << ENDL
<< ENDL
<< "Use \"help\" command to see the list of available commands." << ENDL
<< "**********************************************************************");
m_core.on_synchronized();
}
return true;
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
size_t t_cryptonote_protocol_handler<t_core>::get_synchronizing_connections_count()
{
size_t count = 0;
2014-03-20 11:46:11 +00:00
m_p2p->for_each_connection([&](cryptonote_connection_context& context, nodetool::peerid_type peer_id)->bool{
2014-03-03 22:07:58 +00:00
if(context.m_state == cryptonote_connection_context::state_synchronizing)
++count;
return true;
});
return count;
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
int t_cryptonote_protocol_handler<t_core>::handle_response_chain_entry(int command, NOTIFY_RESPONSE_CHAIN_ENTRY::request& arg, cryptonote_connection_context& context)
{
LOG_PRINT_CCONTEXT_L2("NOTIFY_RESPONSE_CHAIN_ENTRY: m_block_ids.size()=" << arg.m_block_ids.size()
<< ", m_start_height=" << arg.start_height << ", m_total_height=" << arg.total_height);
if(!arg.m_block_ids.size())
{
LOG_ERROR_CCONTEXT("sent empty m_block_ids, dropping connection");
m_p2p->drop_connection(context);
return 1;
}
if(!m_core.have_block(arg.m_block_ids.front()))
{
LOG_ERROR_CCONTEXT("sent m_block_ids starting from unknown id: "
2014-05-25 17:06:40 +00:00
<< epee::string_tools::pod_to_hex(arg.m_block_ids.front()) << " , dropping connection");
2014-03-03 22:07:58 +00:00
m_p2p->drop_connection(context);
return 1;
}
context.m_remote_blockchain_height = arg.total_height;
context.m_last_response_height = arg.start_height + arg.m_block_ids.size()-1;
if(context.m_last_response_height > context.m_remote_blockchain_height)
{
LOG_ERROR_CCONTEXT("sent wrong NOTIFY_RESPONSE_CHAIN_ENTRY, with \r\nm_total_height=" << arg.total_height
<< "\r\nm_start_height=" << arg.start_height
<< "\r\nm_block_ids.size()=" << arg.m_block_ids.size());
m_p2p->drop_connection(context);
}
BOOST_FOREACH(auto& bl_id, arg.m_block_ids)
{
if(!m_core.have_block(bl_id))
context.m_needed_objects.push_back(bl_id);
}
request_missing_objects(context, false);
return 1;
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
bool t_cryptonote_protocol_handler<t_core>::relay_block(NOTIFY_NEW_BLOCK::request& arg, cryptonote_connection_context& exclude_context)
{
return relay_post_notify<NOTIFY_NEW_BLOCK>(arg, exclude_context);
}
//------------------------------------------------------------------------------------------------------------------------
template<class t_core>
bool t_cryptonote_protocol_handler<t_core>::relay_transactions(NOTIFY_NEW_TRANSACTIONS::request& arg, cryptonote_connection_context& exclude_context)
{
return relay_post_notify<NOTIFY_NEW_TRANSACTIONS>(arg, exclude_context);
}
2015-01-05 19:30:17 +00:00
/// @deprecated
template<class t_core> std::ofstream& t_cryptonote_protocol_handler<t_core>::get_logreq() const {
static std::ofstream * logreq=NULL;
if (!logreq) {
LOG_PRINT_RED("LOG OPENED",LOG_LEVEL_0);
logreq = new std::ofstream("logreq.txt"); // leak mem (singleton)
*logreq << "Opened log" << std::endl;
}
LOG_PRINT_YELLOW("LOG USED",LOG_LEVEL_0);
(*logreq) << "log used" << std::endl;
return *logreq;
}
} // namespace