Con Kolivas
|
3d44a13470
|
Store dead proxy instances to be recycled if possible
|
10 years ago |
Con Kolivas
|
b59760bb4a
|
Disable subproxies as they die, moving them to a dead list instead of trying to reuse them and recruit fresh proxies, disconnecting clients connected to them
|
10 years ago |
Con Kolivas
|
aa153ba742
|
Differentiate pong from other spurious messages from clients
|
10 years ago |
Con Kolivas
|
df1465f275
|
Use msg_id variable for clarity
|
10 years ago |
Con Kolivas
|
8ebd5de420
|
Use json_getdel_int64 helper in the connector
|
10 years ago |
Con Kolivas
|
48f7702bd6
|
Add json_getdel_int64 helper
|
10 years ago |
Con Kolivas
|
ed0d20d216
|
Use json_getdel_int helper in generator
|
10 years ago |
Con Kolivas
|
40b10ee2ba
|
Add helper to set a json integer and then delete its entry from the json
|
10 years ago |
Con Kolivas
|
31a5e3d5ce
|
Delete the subproxy entry in the share after sending it to the proxy send thread
|
10 years ago |
Con Kolivas
|
2d35d0c679
|
Parse reconnect differently for subproxies than master proxies
|
10 years ago |
Con Kolivas
|
c43d89275a
|
Fix various logic errors dealing with upstream proxy disconnects
|
10 years ago |
Con Kolivas
|
338cc49c7e
|
Add verbose notify message
|
10 years ago |
Con Kolivas
|
9edfbc8790
|
Don't send proxy diff till it's set yet
|
10 years ago |
Con Kolivas
|
b13446e387
|
Send proxy subscription to stratifier as soon as we have it
|
10 years ago |
Con Kolivas
|
e287feede0
|
Send only the parent proxy id from the generator to the stratifier
|
10 years ago |
Con Kolivas
|
9b04f81666
|
Set correct variable in update_diff
|
10 years ago |
Con Kolivas
|
f15fbdbc70
|
Receive diff from the upstream proxy and add share submission to the correct psend list
|
10 years ago |
Con Kolivas
|
955bbe6977
|
Use the client's sdata to associate correct share submissions with the upstream proxy
|
10 years ago |
Con Kolivas
|
6a21a67c1e
|
Use client sdata with update_client call
|
10 years ago |
Con Kolivas
|
27faf4d1c4
|
Fix proxy epfd not being set
|
10 years ago |
Con Kolivas
|
319d0943fc
|
Update diff on all stratifier data but send diff only to clients bound to the current subproxy
|
10 years ago |
Con Kolivas
|
bcd499bf31
|
Assign the client subproxy data in new_enonce1
|
10 years ago |
Con Kolivas
|
c872ce9de8
|
Store the proxy and subproxy id only in clients and sort subproxies by subid to avoid confusion
|
10 years ago |
Con Kolivas
|
dc293beb9c
|
Separate delivery of messages by ckpool sdata and per proxy sdata
|
10 years ago |
Con Kolivas
|
dd1dae4ae3
|
Send the relevant sdata to be used by new_enonce1
|
10 years ago |
Con Kolivas
|
7763b49c2e
|
Allow stratifier to choose the best subproxy to attach new users to and request recruitment of more subproxies when the headroom for more clients is low
|
10 years ago |
Con Kolivas
|
6822282672
|
Generate stratifier data per subproxy instead of per proxy since they'll have unique workbases
|
10 years ago |
Con Kolivas
|
d62483fbd2
|
Workbases always exists by the time we reach the point we can set the enonce2varlen
|
10 years ago |
Con Kolivas
|
03712fe2d7
|
Broadcast data according to the sdata it belongs to
|
10 years ago |
Con Kolivas
|
b1f92414c5
|
Store which sdata each client is bound to
|
10 years ago |
Con Kolivas
|
e360a9842b
|
Add the workbase to each unique sdata as required
|
10 years ago |
Con Kolivas
|
ba910251da
|
Choose the sdata to use by subproxy
|
10 years ago |
Con Kolivas
|
1f83be877d
|
Send subproxy id when setting proxy at stratifier and set the current proxy to the subproxy instead of the master proxy
|
10 years ago |
Con Kolivas
|
b77027b09c
|
Create a unique stratifer data structure per subproxy
|
10 years ago |
Con Kolivas
|
d05db855ff
|
Create infrastructure for a list of subproxies per proxy and generating them
|
10 years ago |
Con Kolivas
|
08e47d28cf
|
Give each proxy an enonce1 union and cap the handout of enonce1 values without repeating, assuming we will recruit more proxied connections upstream to compensate
|
10 years ago |
Con Kolivas
|
e986c04c68
|
Fix non proxy mode breakage
|
10 years ago |
Con Kolivas
|
c92cbd6f62
|
Handle submitting upstream shares properly with subproxies having their own share tables
|
10 years ago |
Con Kolivas
|
575ac70de8
|
Store all notifies in the generator in the parent proxy list and extract the subproxy details from share submission to submit upstream to the right connection
|
10 years ago |
Con Kolivas
|
3b40b805bd
|
Store the proxy and subproxy id work is generated from
|
10 years ago |
Con Kolivas
|
abde70a8d9
|
Store the id and subid of the proxy in use by each client, submitting the value with its shares to the generator
|
10 years ago |
Con Kolivas
|
a006dab4ae
|
Do not use master proxy for work, using the first subproxy instance for now, sending the id with it to the stratifier
|
10 years ago |
Con Kolivas
|
04d43b3afb
|
Recruit extra subproxies to ensure at least 42 client headroom, receiving data from correct socket in proxyrecv
|
10 years ago |
Con Kolivas
|
fe4a25f861
|
Create rudimentary subproxy structures, moving proxy receives to epoll in order to receive from multiple connections upstream
|
10 years ago |
Con Kolivas
|
163fc40afb
|
Check all delayed clients for a serviceable one in the connector when we can and consider dropping a client servicing one to not potentially create delayed sends faster than we service them
|
10 years ago |
Con Kolivas
|
8b1b6d45d6
|
fd being invalidated is checked for in wait_write_select so we don't need to handle it twice
|
10 years ago |
Con Kolivas
|
0e7bc51541
|
Check all delayed clients for a serviceable one in the connector when we can and consider dropping a client servicing one to not potentially create delayed sends faster than we service them
|
10 years ago |
Con Kolivas
|
83208ebc72
|
fd being invalidated is checked for in wait_write_select so we don't need to handle it twice
|
10 years ago |
Con Kolivas
|
1c1cb445ef
|
Don't send subscriptions to the stratifier in passthrough mode
|
10 years ago |
Con Kolivas
|
b637efa179
|
Handle other forms of read_socket_line ending after message complete as not a failure
|
10 years ago |