Con Kolivas
|
86e70edfbc
|
Make all child processes ping the parent process at regular intervals and then die if it has disappeared. Do not attempt to send unix messages to dead processes and abort instantly on socket EINTR
|
11 years ago |
|
Con Kolivas
|
10a4b36f31
|
Make displayed stats based on round shares
|
11 years ago |
|
Con Kolivas
|
d8b023350f
|
Don't remove the ram from dead client instances to prevent a dereference
|
11 years ago |
|
Con Kolivas
|
c4f0d57ae8
|
Fix various memory leaks
|
11 years ago |
|
Con Kolivas
|
8073936b46
|
Check for incorrect return value on asprintf
|
11 years ago |
|
Con Kolivas
|
164745af2e
|
Clean up unnecessary jansson build components
|
11 years ago |
|
Con Kolivas
|
d90b4ffddd
|
Check return value of asprintf
|
11 years ago |
|
Con Kolivas
|
77f09c2858
|
Use local jansson in build
|
11 years ago |
|
Con Kolivas
|
74db5bcfb8
|
Import jansson 2.6 locally
|
11 years ago |
|
Con Kolivas
|
56d6649646
|
Fix jobid being wrong on proxy reconnect
|
11 years ago |
|
Con Kolivas
|
bfec616da3
|
Create new proxy entries on the fly with the client reconnect command and switch to it
|
11 years ago |
|
Con Kolivas
|
601fb8cb34
|
Fix off-by-one error
|
11 years ago |
|
Con Kolivas
|
d4e6668865
|
Fix minor uninitialised ram and leak issues
|
11 years ago |
|
Con Kolivas
|
3931e7bee9
|
Avoid realloc being required in send_json_msg
|
11 years ago |
|
Con Kolivas
|
a3a586ee6e
|
Use the retry on EINTR only on unix sockets
|
11 years ago |
|
Con Kolivas
|
306392950f
|
Zero the read buffer fully in read_socket_line and handle closed sockets gracefully
|
11 years ago |
|
Con Kolivas
|
67fefd2675
|
Fall over to another bitcoind if any of our calls fail
|
11 years ago |
|
Con Kolivas
|
0b4c6c7821
|
Make the gbtbase data the private data per server instance
|
11 years ago |
|
Con Kolivas
|
e959ba39a4
|
Iterate over the configured btcds, finding the first working one; failover to be added later
|
11 years ago |
|
Con Kolivas
|
82fe18554f
|
Add total shares from each user to pool accounted diff shares
|
11 years ago |
|
Con Kolivas
|
ca176c8e1d
|
Fall over to the first working stratum proxy whenever we are disconnected
|
11 years ago |
|
Con Kolivas
|
3c70130ae2
|
Only show user share summary on block solve if they had any
|
11 years ago |
|
Con Kolivas
|
9b95920ab5
|
Remove redundant comment and there is no point capping shares to 5xdiff
|
11 years ago |
|
Con Kolivas
|
d1e0072efd
|
Load user stats on startup to ensure we don't miss clients that don't reconnect on a pool restart
|
11 years ago |
|
Con Kolivas
|
40807602c9
|
Fix macro variable name clashing and crash on logging prior to logfile being set
|
11 years ago |
|
Con Kolivas
|
7fabe1f14f
|
With sigpipe masked, interrupted send calls are likely fatal
|
11 years ago |
|
Con Kolivas
|
01d5b71bef
|
Broadcast to miners a message when the pool has solved a block
|
11 years ago |
|
Con Kolivas
|
b4e6aef470
|
Log all output to a logfile, display all warnings, and use locking to prevent overlapped messages
|
11 years ago |
|
Con Kolivas
|
f48558d2b5
|
Detach child processes from the tty leaving only the main process to talk to it and exit all processes in a consistent way with more information on termination
|
11 years ago |
|
ckolivas
|
ed270c92fc
|
PPLNS window should just be network diff
|
11 years ago |
|
ckolivas
|
2358d03037
|
Add pplns shares from reconnected users to the pool's total
|
11 years ago |
|
ckolivas
|
d19a682d92
|
Set the maximum pplns window to 5xdiff
|
11 years ago |
|
ckolivas
|
d86b6833ef
|
Retain a proportion of shares across block solves when total shares is greater than the 1/5th window
|
11 years ago |
|
Con Kolivas
|
8e099c8dd1
|
Make the pplns window 1/5 of difficulty
|
11 years ago |
|
Con Kolivas
|
3102379119
|
Limit client diff drops to halving to prevent idle periods invalidating the calculations
|
11 years ago |
|
Con Kolivas
|
305165914d
|
Display a rudimentary pplns summary on solving a block to calculate payouts
|
11 years ago |
|
Con Kolivas
|
1624aaa897
|
Display the pplns shares per worker and store it in the temporary logs
|
11 years ago |
|
Con Kolivas
|
ce028d0485
|
Load the saved pplns shares on beginning a new user instance
|
11 years ago |
|
Con Kolivas
|
f3c899e09b
|
Store a rudimentary summary of pplns shares up to the current network diff in a logfile per user
|
11 years ago |
|
Con Kolivas
|
24643e9ba0
|
Get the network difficulty from the block header and use that to determine possible block solves in proxy mode
|
11 years ago |
|
Con Kolivas
|
ed8f73a426
|
Unix message failures are almost always fatal so add debugging about where the functions were called from
|
11 years ago |
|
Con Kolivas
|
e5953833a5
|
Don't show block hash in logs when accepted
|
11 years ago |
|
Con Kolivas
|
0471340fdd
|
Update notify when we have a new subscription and gracefully drop connections from the send process to be detected rapidly in the recv thread
|
11 years ago |
|
Con Kolivas
|
a50bb29d21
|
Free up old data in place if we are re-subscribing
|
11 years ago |
|
Con Kolivas
|
9983c96ca0
|
Fix uninitialised variable
|
11 years ago |
|
Con Kolivas
|
740edcc80c
|
Revert "Do not hand out enonce1 to new clients when no notify data exists"
This reverts commit cc791f0b67.
|
11 years ago |
|
Con Kolivas
|
5b7af2f7ff
|
Avoid accessing notify instances outside of the notify lock and avoid leaking memory when we can't match the msgid in proxy_send
|
11 years ago |
|
Con Kolivas
|
c9f550207f
|
Send one on all process failure codes
|
11 years ago |
|
Con Kolivas
|
cc791f0b67
|
Do not hand out enonce1 to new clients when no notify data exists
|
11 years ago |
|
Con Kolivas
|
923a4abe59
|
Don't bother sending the listener the shutdown message since the signal handler will kill off the main thread
|
11 years ago |
