Con Kolivas
|
b7186cfcd9
|
Use cached instance variable
|
10 years ago |
Con Kolivas
|
bb8593bfa0
|
Take the same attendant precautions when deleting passthrough client ram as regular clients by adding them to the dead list
|
10 years ago |
Con Kolivas
|
712d5c38d6
|
Make sure a passthrough client is still in the hash table before deleting it
|
10 years ago |
Con Kolivas
|
3a40311ece
|
Passthrough clients should be rare so increase verbosity of them being added
|
10 years ago |
Con Kolivas
|
d6d81093be
|
Fix impossible warnings in drop_client
|
10 years ago |
Con Kolivas
|
9d06777724
|
Do not attempt to decref a freed passthrough client
|
10 years ago |
Con Kolivas
|
b97132dde3
|
Zero wb->logdir even if we don't write to the share log since we still use the value of it
|
10 years ago |
Con Kolivas
|
9892f3af64
|
Use validhex before checking sessionid in disconnected_sessionid_exists to avoid spewing warnings for client errors
|
10 years ago |
Con Kolivas
|
bf1ac3b57b
|
Age disconnected instances over 10 minutes old to the dead list to be reaped
|
10 years ago |
Con Kolivas
|
99dc8947d2
|
Add message about lost ram in block solve and reject.
|
10 years ago |
Con Kolivas
|
1b5e73e57b
|
Fix issues with freeing ram safely on dropping clients and converting them from disconnected to dead clients by removing their instance every time and freeing dead client ram safely when it will no longer be used by the loop.
|
10 years ago |
Con Kolivas
|
cfe29a1e92
|
Fix stratum resume support to work with any supported enonce1 sizes
|
10 years ago |
ckolivas
|
36d814d484
|
Add clients to dead list when dropping all clients instead of losing all references to them
|
10 years ago |
ckolivas
|
17e5f996d8
|
Reinstate add idle, disconnected and dead counts to pool stats
|
10 years ago |
ckolivas
|
ced22bd5ea
|
Only lock/unlock once in drop_client
|
10 years ago |
Con Kolivas
|
b3d251638f
|
Only add authorised clients to disconnected list
Conflicts:
src/stratifier.c
|
10 years ago |
Con Kolivas
|
98bf62ab67
|
Fill enonce1 data properly for stratum reconnect support
|
10 years ago |
Con Kolivas
|
407f087c63
|
Give an outright fail response if failed msg is received from ckdb on auth
|
10 years ago |
Con Kolivas
|
451256f6cb
|
Only give unparseable auth message when failed message is not recognised
|
10 years ago |
Con Kolivas
|
c7600f281b
|
Differentiate generated authed users from unauthorised and don't do any stats on unauthed users
|
10 years ago |
Con Kolivas
|
8c6e786cef
|
Check for corrupt ckdb responses in send_recv_auth
|
10 years ago |
ckolivas
|
2231a38455
|
Read off best share stats if they exist per user and worker
|
10 years ago |
ckolivas
|
62117bb380
|
Reset all best share counters after a block solve
|
10 years ago |
ckolivas
|
5cff5ae477
|
Drop two potentally unhandled reference counts
|
10 years ago |
Con Kolivas
|
d64a0fa5de
|
Dead instances are a singly linked list only
Conflicts:
src/stratifier.c
|
10 years ago |
Con Kolivas
|
4d72f49061
|
Preauth workers for the first 10 minutes after a user first authorises
|
10 years ago |
Con Kolivas
|
8964743cd1
|
Disable resume support till dereference issue in disconnected_instances resolved
|
10 years ago |
Con Kolivas
|
9b56c0c546
|
Fill enonce1 data properly for stratum reconnect support
|
10 years ago |
Con Kolivas
|
b1f7cd833d
|
Give an outright fail response if failed msg is received from ckdb on auth
|
10 years ago |
Con Kolivas
|
2980ff4032
|
Only give unparseable auth message when failed message is not recognised
|
10 years ago |
Con Kolivas
|
9edf91d17e
|
Differentiate generated authed users from unauthorised and don't do any stats on unauthed users
|
10 years ago |
Con Kolivas
|
84dc0146e5
|
Check for corrupt ckdb responses in send_recv_auth
|
10 years ago |
ckolivas
|
0c51e7e863
|
Read off best share stats if they exist per user and worker
|
10 years ago |
ckolivas
|
85230c270e
|
Reset all best share counters after a block solve
|
10 years ago |
ckolivas
|
e1514cdc01
|
Drop two potentally unhandled reference counts
|
10 years ago |
ckolivas
|
24cb6e0c3e
|
Count the number of workernames and use that as the upper limit for the sanity check in statsupdate
|
10 years ago |
ckolivas
|
137203c01a
|
Count each idle client as an idle worker instead of per workername
|
10 years ago |
ckolivas
|
9c262b7fe8
|
Keep worker instances in the linked list since we don't know if we're removing the last one or not
|
10 years ago |
ckolivas
|
e91ea6c235
|
Add sanity breakout of worker instance looping in statsupdate
|
10 years ago |
ckolivas
|
1465002ecc
|
Only lock and release the write instance lock once in drop_client
|
10 years ago |
ckolivas
|
b172162f27
|
kill_instance always follows del_disconnected
|
10 years ago |
ckolivas
|
11fc2fb67d
|
Revert "Only iterate over workers in workerstats if some are left in the worker_instances list"
This reverts commit 450f2040b2 .
Of no advantage.
|
10 years ago |
ckolivas
|
450f2040b2
|
Only iterate over workers in workerstats if some are left in the worker_instances list
|
10 years ago |
Con Kolivas
|
8eee70a4a5
|
Do not delete the worker instance linked list on removing the worker from the list
|
10 years ago |
Con Kolivas
|
33c2465eef
|
Remove worker instances and their data on killing their client instances
|
10 years ago |
Con Kolivas
|
4662873d70
|
Only add authorised clients to disconnected list
|
10 years ago |
Con Kolivas
|
0f56ed8c38
|
Add idle, disconnected and dead counts to pool stats
|
10 years ago |
Con Kolivas
|
9df5a9cf2e
|
Get rid of cute locking in disconnected session check for simplicity
|
10 years ago |
Con Kolivas
|
8f10347775
|
Dead instances are a singly linked list only
|
10 years ago |
Con Kolivas
|
a90b1fd024
|
Preauth workers for the first 10 minutes after a user first authorises
|
10 years ago |