Con Kolivas
|
77e68e8db1
|
Revert "Send reconnect to clients immediately upon detecting they're sending to a dead proxy instead of flagging them"
This reverts commit ac8bb3259d.
|
10 years ago |
|
Con Kolivas
|
0d39d22c16
|
Use the disable_subproxy function for all proxies including the parent and ensure we message the stratifier about all dead proxies, leaving recruitment only to on demand
|
10 years ago |
|
Con Kolivas
|
0afbf3eff2
|
Only consider a proxy dead if all subproxy connections are also dead
|
10 years ago |
|
Con Kolivas
|
ac8bb3259d
|
Send reconnect to clients immediately upon detecting they're sending to a dead proxy instead of flagging them
|
10 years ago |
|
Con Kolivas
|
6971bf45cd
|
Remove the old pid file per process when preparing the new child processes
|
10 years ago |
|
Con Kolivas
|
1bdf57e21d
|
Cope with unknown pids in various send msg commands without terminal failure
|
10 years ago |
|
Con Kolivas
|
0a165bfc66
|
Revert to lazily reconnecting all clients to prevent floods
|
10 years ago |
|
Con Kolivas
|
396b656c82
|
Use the correct json
|
10 years ago |
|
Con Kolivas
|
8dbf39dae5
|
Reconnect clients that are sending shares that can't be processed
|
10 years ago |
|
Con Kolivas
|
a470ca6431
|
Only reconnect clients when there is room for them on the new proxy, tagging the rest and recruiting as needed
|
10 years ago |
|
Con Kolivas
|
08d8644d1a
|
Still drop clients from dead proxies even if we don't have their data any more
|
10 years ago |
|
Con Kolivas
|
73e6824c00
|
Recruit more proxies when we have flagged clients in reconnect_clients
|
10 years ago |
|
Con Kolivas
|
cffc412427
|
Combine the reconnect proxy functions, switching as many clients as there is headroom and flagging the rest for lazy reconnection
|
10 years ago |
|
Con Kolivas
|
51337105ac
|
Count existing reconnects in headroom count
|
10 years ago |
|
Con Kolivas
|
0c185fbd2a
|
Max sure max headroom variable is 64 bit
|
10 years ago |
|
Con Kolivas
|
765c3a050d
|
Reset the pi pid after a failure to find the process alive so we can look it up again in case it has changed
|
10 years ago |
|
Con Kolivas
|
9e5d725e68
|
Make all one way send_procs asynchronous to avoid message response deadlocks
|
10 years ago |
|
Con Kolivas
|
b0060079db
|
Return value of send_proc is never used
|
10 years ago |
|
Con Kolivas
|
dffc938519
|
Free buffer used in generator
|
10 years ago |
|
Con Kolivas
|
9842659e39
|
Fix buf dereference error
|
10 years ago |
|
Con Kolivas
|
47fa24dca0
|
Move reap messages out of lock
|
10 years ago |
|
Con Kolivas
|
f37279da12
|
Check for correct condition in wait_close
|
10 years ago |
|
Con Kolivas
|
7fb37b7886
|
Do not loop after updating immediately in the strat loop
|
10 years ago |
|
Con Kolivas
|
cef4a2cd36
|
Show message associated with no close fd detection
|
10 years ago |
|
Con Kolivas
|
235cdf3c96
|
Flag proxies as dead if they fail connect_proxy
|
10 years ago |
|
Con Kolivas
|
1ff651971f
|
Close socket before adding message to recvs
|
10 years ago |
|
Con Kolivas
|
65d9abab0b
|
Skip subproxies that don't yet have a current workbase when selecting a subscription
|
10 years ago |
|
Con Kolivas
|
111a10bf9d
|
Send clients a reconnect even if we're switching to a backup pool
|
10 years ago |
|
Con Kolivas
|
4cd554a238
|
Set the reconnect time on a proxy when a read socket detects a disconnection
|
10 years ago |
|
Con Kolivas
|
8c07e6eda7
|
Give extra info for extranonce size in update subscribe
|
10 years ago |
|
Con Kolivas
|
ad5b2a089b
|
Add warning about reason for failed subscription
|
10 years ago |
|
Con Kolivas
|
2055f0e2e8
|
Demote nonce length warning for subproxies
|
10 years ago |
|
Con Kolivas
|
41f042c3e5
|
Duplicate the actual jobid json object to return when submitting shares and use an integer consistently as the internal jobid
|
10 years ago |
|
Con Kolivas
|
33f8c4275d
|
Close our end of the socket in send_proc
|
10 years ago |
|
Con Kolivas
|
8dd1754426
|
Close socket reliably in generator
|
10 years ago |
|
Con Kolivas
|
9de62bdea9
|
Reverse the waiting order between generator and stratifier in proxy mode
|
10 years ago |
|
Con Kolivas
|
33e195511c
|
Wait for the other end to close a unix socket to ensure the message has gone through
|
10 years ago |
|
Con Kolivas
|
ea56de3bf6
|
Check for pid in send_recv_proc as well
|
10 years ago |
|
Con Kolivas
|
6e9901f9fe
|
Shutdown instead of closing a socket after sending a unix message allowing the receiving end to close the socket after receiving the data
|
10 years ago |
|
Con Kolivas
|
5a7febc821
|
Workbase id and blockchange id are determined by the generator in proxy mode and should not be set per proxy
|
10 years ago |
|
Con Kolivas
|
29ebae7410
|
Fix incorrect cs being used in proxy_send
|
10 years ago |
|
Con Kolivas
|
68da0e07fa
|
Reap any retired proxies that no longer have any subproxies
|
10 years ago |
|
Con Kolivas
|
e9af372a56
|
Reap old subproxies when they're no longer in use
|
10 years ago |
|
Con Kolivas
|
631467ff70
|
Use a linked list for retired proxies in case they have the same id as a previously retired one
|
10 years ago |
|
Con Kolivas
|
c093b4a06f
|
Keep track of the number of clients bound to each proxy
|
10 years ago |
|
Con Kolivas
|
42f14a9536
|
Add full lock tracking for mutexes and rwlocks
|
10 years ago |
|
Con Kolivas
|
7b452b361f
|
Remove macro placeholders for locks in preparation for implementing lock tracking
|
10 years ago |
|
Con Kolivas
|
865cd3a741
|
Remove unused variable
|
10 years ago |
|
Con Kolivas
|
c572d580eb
|
Add lock contention testing warning at 10 seconds to all lock grabbing and fail if a lock is unable to be grabbed for more than a minute
|
10 years ago |
|
Con Kolivas
|
0f08e9518f
|
Add helpers for rw timed locks
|
10 years ago |
