Con Kolivas
|
96aa070652
|
Revert "Further thread the next two biggest CPU users"
This reverts commit 1789d2b663.
Dangerous.
|
9 years ago |
|
Con Kolivas
|
0b0dfe4f83
|
Move json decoding to threaded client_message_processor
|
9 years ago |
|
Con Kolivas
|
1789d2b663
|
Further thread the next two biggest CPU users
|
9 years ago |
|
Con Kolivas
|
5c09f0608f
|
Microoptimise ckdbq_process
|
9 years ago |
|
Con Kolivas
|
6f8ac8aa97
|
Allow mindiff to be set to zero to reset to defaults with ckdb commands
|
9 years ago |
kanoi
|
a4269d3126
|
ckdb - remove incorrect gcc warning
|
9 years ago |
|
kanoi
|
bd3141462e
|
ckdb - ignore checking the web Anon username for ovents - just check the IP - 2
|
9 years ago |
|
kanoi
|
b1ed1ee158
|
ckdb - ignore checking the web Anon username for ovents - just check the IP
|
9 years ago |
|
kanoi
|
eadb28c76d
|
ckdb - implement ovents
|
9 years ago |
|
kanoi
|
61cb60c00a
|
ckdb - add ok, unok and expire to cmd_events
|
9 years ago |
|
kanoi
|
e0209970e9
|
ckdb - add an enabled flag to events_limits
|
9 years ago |
|
kanoi
|
ddeed87874
|
ckdb - handle events vs ovents
|
9 years ago |
|
Con Kolivas
|
0f5fc8ea8b
|
Use fresh event structure after handing it off to the client event processor
|
9 years ago |
|
kanoi
|
5f4bbd74e4
|
ckdb - add OK events (OVENTS) to allow limiting valid accesses
|
9 years ago |
|
Con Kolivas
|
75923f6daa
|
Make client epoll event handling scalable ckmsgq threads
|
9 years ago |
|
Con Kolivas
|
f0e07c24a4
|
Make client event handling a oneshot event that is rearmed to allow us to thread the work
|
9 years ago |
|
Con Kolivas
|
39070094cf
|
Invalidate clients in common location after parse_client_msg
|
9 years ago |
|
Con Kolivas
|
ebedb77629
|
Cache reused variables in creceiver
|
9 years ago |
|
kanoi
|
465ab648f8
|
ckdb - use optioncontrol for storing permanent IPS OK and bans
|
9 years ago |
|
kanoi
|
22292423e8
|
ckdb - add cmd_events for managing/viewing event info
|
9 years ago |
|
kanoi
|
224de3c5fe
|
ckdb - add triggers for updating global varaibles from optioncontrol
|
9 years ago |
|
Con Kolivas
|
db42a137e1
|
Implement a ckdbflush message for emergency use only
|
9 years ago |
|
Con Kolivas
|
0cb8d0af48
|
Fix client buffer leak in connector
|
9 years ago |
|
Con Kolivas
|
f89ace7a3d
|
Silence uninitialised byte warnings in valgrind
|
9 years ago |
|
Con Kolivas
|
8007f6524b
|
Writes blocking in send_sender_send are not an unlikely event
|
9 years ago |
|
Con Kolivas
|
6986ba95fc
|
Make client message processing a ckmsq to minimise connector_loop holdup
|
9 years ago |
|
kanoi
|
09a40a4607
|
ckdb - add some table level locking to speed up large loads
|
9 years ago |
|
kanoi
|
3aff67018f
|
ckdb - allow blocking and script alerting on ckdb events
|
9 years ago |
|
Con Kolivas
|
df346876d8
|
Only send node workinfo in pool mode
|
9 years ago |
|
Con Kolivas
|
2a76d1ab53
|
Add a failsafe mode to nodes that will allow miners to continue hashing even if workinfo is temporarily unavailable
|
9 years ago |
|
Con Kolivas
|
6f48d76e7e
|
Tidy up transaction refcount by resetting it every time it's seen instead of just incrementing it, differentiating node refcounts from main pools
|
9 years ago |
|
Con Kolivas
|
089b0440c1
|
Increase refcount when sent a duplicate transaction on nodes
|
9 years ago |
|
Con Kolivas
|
1fa449be67
|
Temporary workaround making nodes match current pool behaviour with txn refcount increasing by 2 each time they're accessed
|
9 years ago |
|
Con Kolivas
|
b046e5c01b
|
Don't increment the refcounter when adding transactions on the main pool to avoid getting ahead of nodes and set default refcount to 10 minutes backlog of 20 workinfos
|
9 years ago |
|
Con Kolivas
|
0b8892db00
|
Increase txn backlog to nodes to 100
|
9 years ago |
|
Con Kolivas
|
67d6a5ee6d
|
Don't stack low priority updates
|
9 years ago |
|
Con Kolivas
|
b6588c5f13
|
Implement a cksem_trywait function
|
9 years ago |
|
Con Kolivas
|
f7bf3592d8
|
Tidy up node transaction notices
|
9 years ago |
|
Con Kolivas
|
df11fda8ea
|
Reset failed_authtime on successful authorisation
|
9 years ago |
|
Con Kolivas
|
2c6b550732
|
Test if there is a btcaddress every time a user is generated in case a previous call to test_address failed
|
9 years ago |
|
Con Kolivas
|
4df76f0646
|
Revert "Revert "Add transactions to a hashtable to know which are new, which need propagating, and which are no longer needed""
This reverts commit 147175f295.
Reinstate for merge.
|
9 years ago |
|
Con Kolivas
|
3a7ea5b504
|
Revert "Revert "Move merkle tree generation to the stratifier along with all transaction data to allow transaction data to be used directly in future code""
This reverts commit d0398fed4a.
Reinstate for merge.
|
9 years ago |
|
Con Kolivas
|
2c0b03877b
|
Create the ckdb heartbeat thread to keep throbber moving even when standalone pool/node is idle
|
9 years ago |
|
Con Kolivas
|
71f0077f74
|
Make new node compatible with old pool comms protocol
|
9 years ago |
|
Con Kolivas
|
76dea2d7fa
|
Use the existing txn_hashes string to rebuild transactions on nodes
|
9 years ago |
|
Con Kolivas
|
8d3dbb790f
|
Array transaction order matters and json_array_foreach does not iterate in order
|
9 years ago |
|
Con Kolivas
|
b826ac25ab
|
Purge old transactions on mining nodes whose refcount drops to zero
|
9 years ago |
|
Con Kolivas
|
0b7e3994b2
|
Rebuild workinfo from transaction hashes on nodes
|
9 years ago |
|
Con Kolivas
|
8e76983fa8
|
Send all transactions to new nodes, and set initial refcount on transactions to 10
|
9 years ago |
|
Con Kolivas
|
41a9788cc9
|
Add transactions when received by nodes to the txntable
|
9 years ago |
