RanchiMall/electrumx
3
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
622 Commits 4 Branches 138 Tags 3.9 MiB
5fe49bb261
Commit Graph

119 Commits

Author SHA1 Message Date
Neil Booth
5fe49bb261 Synchronize daemon height and mempool fetching 
Cleanup and simplify touched handling and its event, which is
now controlled and owned by the mempool.
The daemon object owns the set of current mempool hashes.
Clean up and simplify the mempool main loop.

Fixes #70.
 2016-12-15 12:08:14 +09:00
Neil Booth
e2f4847632 New env var MAX_SESSIONS 
When the number of sessions reaches MAX_SESSIONS, which defaults
to 1,000, turn off TCP and SSL listening sockets to prevent new
connections.  When the session count falls below a low watermark,
currently 90% of MAX_SESSIONS, the listening sockets will be re-opened.
Helps prevent DoS and limit open file usage.

Bug fix: do not start serving paused connections until the buffer
socket is sufficiently drained.  Also, loop.
 2016-12-13 22:47:03 +09:00
Neil Booth
1ef6a4d785 Split mempool out into new file. 
Rework mempool handling
 2016-12-12 00:05:42 +09:00
Neil Booth
7399452294 Temporary hack to fix initial sync slowdown 2016-12-08 22:47:29 +09:00
Neil Booth
238b9e0285 Process blocks in the executor 
This should eliminate latency when new blocks come in
and prevent sessions being dropped whilst they are processed.
 2016-12-03 23:32:20 +09:00
Neil Booth
6b551973e8 queue -> tasks 2016-11-29 23:21:22 +09:00
Neil Booth
f9cc21807f Further optimize the inner loop 2016-11-27 14:20:36 +09:00
Neil Booth
5c0b026158 Tasks queue just has null entries for now 2016-11-27 13:35:09 +09:00
Neil Booth
6aef79461f Don't shut down block processor by cancellation 
The block processor needs to be able to close cleanly, and not
mid-block.  In order to be able to yield whilst processing blocks
we cannot forcefully close its coroutine with a cancellation.
 2016-11-27 12:20:16 +09:00
Neil Booth
1b95bcd8ac Open DB differently depending on if syncing 
If syncing, use a high max_open_files, otherwise lower it.
 2016-11-27 07:25:28 +09:00
Neil Booth
0afddb7bc3 Give backing up its own flush function 
Remove some excessive log messages
 2016-11-26 20:59:59 +09:00
Neil Booth
f5384ccc98 Remove get_tx_hash 
Anything in the DB should be on the disk.  Remove misleading
comment.
 2016-11-26 20:09:18 +09:00
Neil Booth
3f743e9630 Distinguish undo_info being empty and not existing 
Fixes #46
 2016-11-26 16:16:17 +09:00
Neil Booth
af3ae6429e Fix logging bug 2016-11-26 15:35:55 +09:00
Neil Booth
95a2de6dae Move DB undo code to db.py 2016-11-26 13:18:10 +09:00
Neil Booth
8d253c81d3 Move clean_db() to db.py 2016-11-26 13:15:45 +09:00
Neil Booth
db5d0dd6cb Merge branch 'serve_disk_only' into develop 2016-11-26 13:12:57 +09:00
Neil Booth
59244cc24e Move bulk of FS flush code to db.py 
More logically belongs there.
Have servers use the flushed DB height not the block processor's height.
Get all headers only from disk.
 2016-11-26 13:11:59 +09:00
Neil Booth
df3cb8535a Cleaner handling of touched addresses 
Moved to local vars rather than an instance variable
 2016-11-26 12:46:11 +09:00
TheLazieR Yip
f3ecfe00e2 Remove unused variables 2016-11-25 07:45:45 -05:00
TheLazieR Yip
828727d41a Remove unneccessary local variables 2016-11-25 07:44:19 -05:00
TheLazieR Yip
b3623f5455 replace header_hashes with header_prevhash , header_hash 2016-11-25 07:25:56 -05:00
Neil Booth
ef2703177e Fix another assertion error during reorgs 
Fixes #44
 2016-11-25 08:10:11 +09:00
Neil Booth
9e220820aa Rework futures and event handling 2016-11-23 17:52:09 +09:00
Neil Booth
bab0d162de Merge ServerManager and BlockServer 2016-11-23 17:52:08 +09:00
Neil Booth
640360c809 Add feature to simulate reorgs for debugging 2016-11-23 09:52:30 +09:00
Neil Booth
8970205e6c Remove obsolete debugging feature 2016-11-23 09:16:41 +09:00
Neil Booth
2df5aa746f Pop one from tx_counts for each block we back up. 
Fixes #40
 2016-11-23 09:11:49 +09:00
Neil Booth
26221e751e Remove dead code 2016-11-23 09:02:01 +09:00
Neil Booth
d1ba4b1a76 Reset fs_height when backing up 2016-11-22 08:09:43 +09:00
Neil Booth
67bb01bf2c Fix reorgs with unspendable TXOs 2016-11-22 07:50:01 +09:00
Neil Booth
87cdd2709d Implement daemon failover 
Daemon URLs can be comma-separated in the DAEMON_URL env var.
Surrounding whitespace is stripped.
http:// is preprended if missing.
The coin's default port is supplied if missing.
A trailing / is supplied if missing.
Closes #33
 2016-11-20 20:52:54 +09:00
Neil Booth
98fd178c62 Make mempool truly asynchronous 
Mempool updates happen in parallel asynchronously to processing of
new blocks once caught up.  This means that, e.g., during the initial
slow mempool download incoming blocks can be processed and communicated
to clients without waiting for the downloaded mempool transaction
analysis to complete.  From a client's point of view the server won't
be seen as lagging.
 2016-11-20 18:17:55 +09:00
Neil Booth
8617c82ec2 Floor disk_count at zero. 2016-11-19 20:56:08 +09:00
Neil Booth
52116539d4 Clean shutdown: wait for transports to close 
We give 10 seconds for everything to be cleaned up, then close forcibly.
Fixes #30
 2016-11-19 17:05:47 +09:00
Neil Booth
3d87e299ea Move formatted_time to library 2016-11-19 15:42:06 +09:00
Neil Booth
942d5d6b00 Fix import 2016-11-19 08:07:37 +09:00
Neil Booth
063a5469cc Move mempool to be with BlockServer 
BlockProcessor doesn't really need a mempool; mempools are
only useful for servers.
Set first_sync before flushing, so it goes into DB state.
Start servers immediately on catchup; mempool fills asynchronously.

Fixes #31
 2016-11-19 08:02:13 +09:00
Neil Booth
d6fa4a1564 More logging improvements 
- log software and DB version at startup
- log more DB info at startup
- log software and DB version when first synced

Fixes #29
 2016-11-19 07:29:17 +09:00
Neil Booth
88185b0daa Bump log threshold up; tweak log 2016-11-18 22:20:48 +09:00
Neil Booth
24d9e5a95e Tweak flush logic 
Aim to not check cache size if caught up; also reduces logs
 2016-11-18 22:17:34 +09:00
Neil Booth
e6058c3fd3 hex_hash is already hex 2016-11-18 21:16:34 +09:00
Neil Booth
d2c575a24b Better, more concise logs, particularly when caught up. 2016-11-18 20:08:43 +09:00
Neil Booth
3f8b26291a At startup only log some things during initial sync 2016-11-18 19:21:45 +09:00
Neil Booth
f4651c0ca4 Store tx_num in 'h' table key so keys are unique 2016-11-17 07:59:40 +09:00
Neil Booth
3c98053f5d Remove some dead code 2016-11-17 07:44:42 +09:00
Neil Booth
d85034353f Remove useless re-raise of cancellation 2016-11-15 20:13:50 +09:00
Neil Booth
db187540d5 Have task coros handle cancelled exceptions 2016-11-15 20:09:38 +09:00
Neil Booth
a2280bbc93 Clean up futures handling 2016-11-15 20:09:38 +09:00
Neil Booth
8671e57183 Move fs flushes to same time as history flushes 2016-11-15 06:28:52 +09:00