Commit graph

117 commits

Author SHA1 Message Date
Howard Chu
2b0fa05f0d Another take on migration
Delete old indices and recreate them, rather than updating them
Maybe not quite as slow as before.
2016-04-24 17:46:50 +01:00
Howard Chu
c14f9efd52 Migration
Migrate from DB version 0 to version 1 on startup
2016-04-08 03:11:05 +01:00
Howard Chu
d7ea7d9a23 Merge branch 'performance' into master 2016-04-05 21:13:16 +01:00
Howard Chu
372acee723 Cleanup
drop obsolete remove_output()
fix get_output_key(global), fix crash in blockchain_dump
2016-04-05 21:05:24 +01:00
Howard Chu
591e421875 Cleanup and clarify
Try to rationalize the variable names, document usage.
2016-04-05 20:57:45 +01:00
Howard Chu
b2f1c58805 Use cursors in some remove functions
Helps when they're called repeatedly in one txn
2016-04-05 20:57:31 +01:00
Howard Chu
118dd69dd5 Use DUPFIXED for block_info and output_txs
Saves another ~150MB or so on the full blockchain
2016-04-05 20:55:16 +01:00
Howard Chu
6225716f3c More outputs consolidation
Also bumped DB VERSION to 1
Another significant speedup and space savings:
Get rid of global_output_indices, remove indirection from output to keys

This is the change warptangent described on irc but never got to finish.
2016-04-05 20:55:12 +01:00
Howard Chu
7c5abdc3a3 Use DUPFIXED for output_keys
Saves another 90MB on 200000 block import.
Had to bring back compare_uint64 for this, but it's safe since
this table is always 64-bit aligned.
2016-04-05 20:54:42 +01:00
Howard Chu
8e9d8e3364 Use DUPFIXED for tx_indices
Small space savings, no measurable speedup
2016-04-05 20:54:39 +01:00
Howard Chu
a12f9365f8 Use DUPFIXED for block_heights
Only a small savings...
2016-04-05 20:54:07 +01:00
Howard Chu
38c2277d6f Use DUPFIXED for spent_keys 2016-04-05 20:54:07 +01:00
warptangent
bf769c32ba Add back changes from revert.
m_tx_outputs doesn't need to be changed, as it's no longer dup list.
2016-04-05 20:54:07 +01:00
warptangent
46b991b362 Use MDB_APPEND mode with two tx subdbs
This is possible on those using a tx index as a key.
2016-04-05 20:54:07 +01:00
warptangent
9aadedb1d0 Schema update: tx_indices - consolidate the tx subdbs from 5 to 3 2016-04-05 20:54:06 +01:00
warptangent
a2f518aa01 Schema update: tx_indices - yet less indirection 2016-04-05 20:54:06 +01:00
warptangent
8d12a8df2c Schema update: tx_indices - improve further with less indirection 2016-04-05 20:54:06 +01:00
warptangent
7c013f66e9 Add batch warning for further review 2016-04-05 20:54:06 +01:00
warptangent
ae0854a431 Schema update: tx_indices 2016-04-05 20:54:06 +01:00
Howard Chu
8d252a4214 Consolidated block info 2016-04-05 20:53:59 +01:00
warptangent
132c666f67 Update schema for "tx_outputs" to use array containing amount output indices
This speeds up wallet refresh by directly retrieving a tx's amount output indices.

It removes the indirection and walking the amount output duplicate list
for every amount in each requested tx.

"tx_outputs" is used by:
Amount output indices are needed for wallet refresh.
Global output indices are needed for removing a tx.

Both amount output indices and global output indices are now stored in
an array of 64-bit unsigned ints:

tx_outputs[<tx_hash>] -> [ <a1_oi, a1_gi, a2_oi, a2_gi, ...> ]

Previously it was:
tx_outputs[<tx_hash>] -> duplicate list of <a1_gi, a2_gi, a3_gi, ...>

The amount output list had to be walked for every amount in order to
find each amount's output index, by comparing the amount's global output
index with each one in the duplicate list until a match was found.

See also d045dfa7ce
2016-04-05 20:30:50 +01:00
Howard Chu
efbdde2c66 Detect map resize failures 2016-03-30 20:37:28 +01:00
moneromooo-monero
600a3cf0c0
New RPC and daemon command to get output histogram
This is a list of existing output amounts along with the number
of outputs of that amount in the blockchain.

The daemon command takes:
- no parameters: all outputs with at least 3 instances
- one parameter: all outputs with at least that many instances
- two parameters: all outputs within that many instances

The default starts at 3 to avoid massive spamming of all dust
outputs in the blockchain, and is the current minimum mixin
requirement.

An optional vector of amounts may be passed, to request
histogram only for those outputs.
2016-03-26 21:10:43 +00:00
moneromooo-monero
79117d4275
db_lmdb: include the error codes from lmdb api in error logs 2016-03-20 18:06:04 +00:00
Howard Chu
db1b2db4d5 Reduce log noise 2016-03-19 12:59:05 +00:00
Howard Chu
a74348e115 Add destructor for readtxns
Only if we created the readtxn. Was missing cleanups from exceptions before.
2016-03-16 11:34:13 +00:00
Howard Chu
6b0a903177 Small cleanups
Only one return and TXN_POSTFIX_RDONLY() per function
Only log rtxn_start if the rtxn wasn't already active
2016-03-15 13:35:31 +00:00
Howard Chu
01c1512f22 More for 92dd4ec6d6
Make sure we stop the right txn too
2016-03-15 13:35:23 +00:00
Riccardo Spagni
240a50f3fb
Merge pull request #723
2abdb2c avoid some val copies (Howard Chu)
2016-03-14 22:20:42 +02:00
Howard Chu
92dd4ec6d6 Hack for read/write txn mixup
save the thread ID of the writer thread so we don't try to use
the writetxn from reader threads
2016-03-14 20:19:46 +00:00
Howard Chu
2abdb2c9fd avoid some val copies 2016-03-14 09:40:49 +00:00
Howard Chu
8941ce0398 More for bdec7cb
More uses of db error helper
2016-03-09 18:24:16 +00:00
Riccardo Spagni
32dc08d19a
Merge pull request #700
bdec7cb BlockchainLMDB: Use DB error helper consistently (warptangent)
c5932eb BlockchainLMDB: Add DB error to exception (warptangent)
a49c355 Blockchain: Omit verbose time stats messages by default (warptangent)
2016-03-05 23:10:38 +02:00
Riccardo Spagni
c3af15702f
Merge pull request #698
ee7a8b8 Get rid of lmdb_cur (Howard Chu)
2016-03-05 23:06:26 +02:00
warptangent
bdec7cbfb3
BlockchainLMDB: Use DB error helper consistently 2016-03-03 21:27:13 -08:00
warptangent
c5932eb5ea BlockchainLMDB: Add DB error to exception 2016-03-03 20:15:37 -08:00
Howard Chu
ee7a8b87f9 Get rid of lmdb_cur
We don't need it now with per-txn cursors.
2016-03-03 19:06:27 +00:00
Howard Chu
f5affbef26 More for df239428c0
Let ARMv7 work again
2016-03-03 00:13:13 +00:00
warptangent
2c823c1589
Make partial revert consistent
If user-defined comparator is used, subdb shouldn't be opened with
MDB_INTEGERKEY.

TODO: Again, this will be added back with future schema updates.
2016-03-01 02:47:16 -08:00
warptangent
df239428c0
Revert part of 7db89ed2ee
For now, so existing databases work.

TODO: add these back with future schema updates.
2016-02-29 18:03:41 -08:00
warptangent
c9c4060dc6
BlockchainLMDB: Add height and db error to exception 2016-02-29 17:52:54 -08:00
Howard Chu
7db89ed2ee ARMv7: fix unaligned accesses
And cleanup some key comparators
2016-02-25 13:57:00 +00:00
Howard Chu
8cc7a36f0b read txn/cursor stuff
Could wrap more later.
2016-02-23 20:47:15 +00:00
Howard Chu
02abe3590d Use MDB_PREV_MULTIPLE
in get_global_output_indices
2016-02-17 16:14:50 +00:00
Howard Chu
7a4755d3a4 Fixup after lmdb master resync 2016-02-17 16:14:38 +00:00
Howard Chu
3b13a7473e Shutup about VERSION 0 2016-02-17 05:06:19 +00:00
Howard Chu
1537477c9f Use cursor in get_output_key 2016-02-17 04:05:29 +00:00
Howard Chu
f2faf8cdd9 Use MDB_APPEND mode where possible
When keys are contiguous and monotonically increasing, this gets
denser page utilization (doesn't leave padding in page splits).
Can't be used for keys that are inserted in random order (e.g. hashes)
In total this only saves around 1.5% of space compared to original
DB code. The previous patch accounted for 0.8% savings on its own;
the blocks tables just aren't that big.
2016-02-17 04:05:29 +00:00
Howard Chu
090b548c3b Use cursors in write txns
Saves a bit of seek overhead. LMDB frees them automatically
in txn_(commit|abort) so they need no cleanup.
2016-02-17 04:05:29 +00:00
Howard Chu
ed08d2152e Keep a running blocksize count
Used in batch size estimation, avoids rereading already processed
blocks during import
2016-02-17 04:05:28 +00:00