JoelKatz
b6871cba8d
Dispatch new pathfinding requests.
2013-05-02 16:13:36 -07:00
JoelKatz
3ec116f5bf
Monitor LevelDB performance.
2013-04-30 17:43:56 -07:00
JoelKatz
965f33e222
Remove log.
2013-04-22 07:35:25 -07:00
JoelKatz
6bbef15e4d
Don't crash on shutdown.
2013-04-21 17:19:41 -07:00
JoelKatz
79ec8e6070
First part of fetch acceleration changes. Includes a bugfix that I'll push to master.
2013-04-21 12:44:21 -07:00
JoelKatz
8215463879
Thread dispatch fixes. (code is commented out)
2013-03-29 18:28:16 -07:00
JoelKatz
4f029e25b6
Temporarily back out the thread migration code until I can figure out if Boost's IO service is broken.
2013-03-29 17:56:03 -07:00
JoelKatz
75dd924f3a
Don't spin when an IO+ thread returns to the job queue.
2013-03-29 17:07:42 -07:00
JoelKatz
9cbd019874
A small change with big consequences. Allow job queue threads to moonlight as I/O threads.
2013-03-29 10:18:11 -07:00
JoelKatz
9ceb380cb8
Make debug easier by naming threads.
2013-03-20 05:41:08 -07:00
JoelKatz
b9456c8fd7
Except when interacting with boost, use std::function and std::bind
...
when C++11 is available. See: http://stackoverflow.com/questions/14617835
2013-03-16 03:27:43 -07:00
JoelKatz
e392366c63
Get ledger acquire work off the main thread and out of the master lock.
...
Prioritize ledger acquire work appripriately. This should help prevent the
server from becoming overly sluggish while acquiring ledgers. Still todo:
Finish all pending receive operations before sending out any data requests.
2013-03-15 22:34:55 -07:00
JoelKatz
e9a02882f1
Don't let ledger fetching go too fast.
2013-03-12 15:17:26 -07:00
JoelKatz
6f40e17bad
Remove broken workaround.
2013-03-10 17:04:10 -07:00
JoelKatz
0007d76923
Job queue simplificaiton
2013-03-06 19:30:08 -08:00
JoelKatz
dc3d82272e
Track threads working on jobs.
2013-03-06 19:07:54 -08:00
JoelKatz
253485e3db
Workaround for whatever the real bug is.
2013-03-06 17:01:15 -08:00
JoelKatz
a0a0a27bbf
Load factor adjustment and reporting tweaks.
2013-03-06 11:20:15 -08:00
JoelKatz
bc99fdc45f
More accurate field names.
2013-03-03 16:27:11 -08:00
JoelKatz
f0c029ef0d
Rework the way load is accounted to be more accurate and more specific.
2013-03-03 16:24:47 -08:00
JoelKatz
c984f83eb6
Fix for job destruction race condition.
2013-03-01 18:27:07 -08:00
JoelKatz
2de9ffe4dc
Up target latencies for client commands.
2013-02-20 14:28:16 -08:00
JoelKatz
cb111b6318
Path finding fixes:
...
Track path finding and order book creation time and latency.
Use a ledger snapshot for path finding.
You can't call getCurrentLedger from an unlocked context.
Pathfinding called from the transaction sign function held the master lock.
2013-02-12 20:39:00 -08:00
JoelKatz
4261b1c2d0
Quick hack to stop core dumps on shutdown. The client needs to be shutdown before
...
the job queue. This just prevents it from tripping.
2013-02-11 07:00:09 -08:00
JoelKatz
4280214a94
Typo.
2013-02-10 16:57:38 -08:00
JoelKatz
a26a302476
Accept old ledgers acquired over the network at a lower priority.
2013-02-10 16:40:52 -08:00
JoelKatz
5aa810404b
Track uptime. Update local fee schedule based on load manager.
2013-02-05 21:33:42 -08:00
JoelKatz
faa344c7d0
Dispatch write ahead logging through our job queue. SQL logging improvements.
2013-01-28 14:32:52 -08:00
JoelKatz
e48ef29f8c
Dispatch HashedObject background writes through our job queue.
2013-01-28 07:13:05 -08:00
JoelKatz
c543b77699
Demote debug.
2013-01-28 01:13:55 -08:00
JoelKatz
80d98e55a9
Clean up JSON reporting of booleans and network state.
2013-01-18 17:38:11 -08:00
JoelKatz
c88c7c50a8
Oops. Fix the job run order.
2013-01-11 18:12:42 -08:00
JoelKatz
db1322f741
The priority scheme was backwards!
2013-01-10 15:29:52 -08:00
JoelKatz
03ecbd1ea8
Create a few extra threads.
2013-01-04 00:22:20 -08:00
JoelKatz
1127ae560e
Defer publishing a ledger to clients until it accumulates sufficient
...
validations. We now have an open ledger, a last closed ledger, and a last
validated ledger. Normally, a ledger will be validated right after it
closes, but in edge cases, we might see a ledger close that then doesn't get
validated. This makes the code do the right thing.
2013-01-02 12:04:16 -08:00
JoelKatz
69a434763e
Fix a bug Arthur reported.
2012-12-24 14:52:36 -08:00
JoelKatz
34456b019c
Cleanup thread creation.
2012-12-21 12:10:26 -08:00
JoelKatz
0106365098
Improve log level selection from command line. Some extra shutdown logging.
2012-12-20 09:27:50 -08:00
JoelKatz
7172946b8d
For now, limit to one transaction thread in standalone mode.
2012-12-17 23:56:13 -08:00
JoelKatz
2a06686b7c
Round one of fixes to avoid ridiculous numbers of spurious copy constructor and destructor calls.
...
Most of these fixes involve calls to BOOST_FOREACH to iterate over a map or unordered_map where the
iterator type didn't perfectly match the internal type, so a reference into the map couldn't be created
and a new value/content pair had to be created for each iteration.
2012-12-17 20:20:24 -08:00
JoelKatz
d046730223
Latency targets.
2012-12-03 04:40:26 -08:00
JoelKatz
8b53f85f60
Separate ledger publish logic so we can tell if clients are overloading us.
2012-11-23 12:15:04 -08:00
JoelKatz
7d4d18bf8d
Add RPC to load tracking.
2012-11-20 13:04:17 -08:00
JoelKatz
5ac22ff31b
Two more load monitoring hooks.
2012-11-20 12:09:51 -08:00
JoelKatz
11088937a4
Track special disk accesses.
2012-11-19 17:08:58 -08:00
JoelKatz
82c4d81b0e
Better naming.
2012-11-19 15:44:54 -08:00
JoelKatz
cec3aaf4b3
Bugfixes.
2012-11-19 15:43:55 -08:00
JoelKatz
19d73bd477
More tie ins.
2012-11-19 15:41:45 -08:00
JoelKatz
08f271443d
Start to hook this stuff up.
2012-11-19 15:33:49 -08:00
Stefan Thomas
fa3fab5816
Moved cpp code to src/cpp and js code to src/js.
2012-11-07 12:49:50 -08:00
