Con Kolivas
|
e8179cd951
|
Drop log level for failed auth json decode failure
|
10 years ago |
Con Kolivas
|
fcfc06e77b
|
More logic fail
|
10 years ago |
Con Kolivas
|
5fd152c370
|
Reverse logic
|
10 years ago |
Con Kolivas
|
f3254ea170
|
Check for addrauth as well in send_recv_auth
|
10 years ago |
Con Kolivas
|
574d028d95
|
Parse the ckdb heartbeat response for setting worker mindiff
|
10 years ago |
Con Kolivas
|
6a72525b8b
|
Search for authorise in response instead of auth to match query
|
10 years ago |
Con Kolivas
|
27ac4e23ef
|
Parse the difficultydefault entry returned from authorise
|
10 years ago |
Con Kolivas
|
3288b15622
|
Parse new auth command from ckdb
|
10 years ago |
Con Kolivas
|
03878efb0e
|
The stats on idle miners were being decayed with the wrong time interval so would appear to drop to almost zero on any idle period
|
10 years ago |
Con Kolivas
|
b08f4bcb69
|
Drop clients lazily when the proxy is full
|
10 years ago |
Con Kolivas
|
1b45079802
|
Keep track of clients that submit a run of continuous rejects and send them a diff update if they do it for longer than 1 minute and then lazily drop them if it goes beyond 2 minutes
|
10 years ago |
Con Kolivas
|
1f68a5a98b
|
Use the Close handler in the stratifier
|
10 years ago |
Con Kolivas
|
070a1a3b3e
|
Cache reused bias calculations
|
10 years ago |
Con Kolivas
|
6acb482e7a
|
Bias workers as well as users
|
10 years ago |
Con Kolivas
|
6ba4cf50a5
|
There is no need to bias 60 second times, and we should bias worker stats
|
10 years ago |
Con Kolivas
|
1e175d851c
|
Missing valid but not submitted count for proxy
|
10 years ago |
Con Kolivas
|
53641a9c24
|
Prevent wb dereference in parse_submit
|
10 years ago |
Con Kolivas
|
600e6923fc
|
Count only accepted and stale rejects in diff calculation
|
10 years ago |
Con Kolivas
|
0f7daecac4
|
Stats on 1st reject were being missed
|
10 years ago |
Con Kolivas
|
255bab88fe
|
Add a maximum diff option
|
10 years ago |
Con Kolivas
|
857462e16a
|
Simplify the complicated sequential locking to pure choice of write or read in the stratifer
|
10 years ago |
Con Kolivas
|
55ff1389ee
|
Make ckpool built without ckdb support imply standalone and remove ckdb options when build support is disabled
|
10 years ago |
Con Kolivas
|
49e465c8d5
|
Remove unnecessary extra step in disconnected_sessionid_exists
|
10 years ago |
Con Kolivas
|
00fc686bbf
|
use a union for the split nonce2 values instead of open coding a pointer version
|
10 years ago |
Con Kolivas
|
ba506f6262
|
Fix more 64bit print modifiers
|
10 years ago |
Con Kolivas
|
82c7f24614
|
Add support for setting worker diff and mining.suggest_difficulty parsing
|
10 years ago |
Con Kolivas
|
fff2028591
|
Avoid uninit buf
|
10 years ago |
Con Kolivas
|
87336835f6
|
Log combined per worker stats in a workers subdirectory
|
10 years ago |
Con Kolivas
|
f54aa412b1
|
Create a linked list of each worker per user and decay per worker shares separately
|
10 years ago |
Con Kolivas
|
dd6a576935
|
Remove the linked list of dead clients and maintain a linked list of all workers by user, and store stats per combined workername
|
10 years ago |
Con Kolivas
|
991877134a
|
Send a heartbeat every second to ckdb to allow ckdb to send information not specifically requested while maintaining ckpool msg-response relationship. Do not send heartbeats if there are messages queued already for when ckdb is offline/busy.
|
10 years ago |
Con Kolivas
|
17c2276e33
|
Ensure we have serialised access to ckdb by using a mutex on submit/receives for auths since other responses are mostly ignored
|
10 years ago |
Con Kolivas
|
e441e65c53
|
Ping miners regularly in proxy mode to ensure they don't disconnect when diff is too high for them and they receive no update
|
10 years ago |
Con Kolivas
|
eeba52e716
|
Remove enonce1 debugging
|
10 years ago |
Con Kolivas
|
8c44df96cb
|
Ignore reject floods on initial connect
|
10 years ago |
Con Kolivas
|
c4506c7e47
|
Round away from zero when choosing optimal diff
|
10 years ago |
Con Kolivas
|
3d6601d20c
|
Maximise the nonce2 size in proxy mode and ensure we don't clash enonce1 instances, rejecting clients when there is no more space available
|
10 years ago |
Con Kolivas
|
36436b0c4c
|
Use singly linked list only for dead clients and don't keep disconnected instances in proxy mode
|
10 years ago |
Con Kolivas
|
79c99c4daf
|
Must check against upstream diff before submitting
|
10 years ago |
Con Kolivas
|
d928c154ee
|
Submit only valid and stale shares upstream in proxy mode and add logging about submits in verbose mode
|
10 years ago |
Con Kolivas
|
10bb351736
|
Avoid updating diff until there is a current work base
|
10 years ago |
Con Kolivas
|
427c1bf12c
|
Check diff on each notify in proxy mode
|
10 years ago |
Con Kolivas
|
f4f6d53db9
|
Submit stale shares in proxy mode
|
10 years ago |
Con Kolivas
|
6d70aac3a2
|
Clamp lower diff to 1 in proxy mode
|
10 years ago |
Con Kolivas
|
08f7f39836
|
Do not accept proxy notify update until subscription complete
|
10 years ago |
Con Kolivas
|
e326d36be5
|
Only use 4 bytes for enonce1 in proxy mode when we have 8 or more bytes for nonce2len
|
10 years ago |
Con Kolivas
|
8e34ff377b
|
Make necessary changes to ckproxy to support large coinbases and null length enonce1 for p2pool support
|
10 years ago |
Con Kolivas
|
b36e4cec3c
|
Make sure jobids align between generator and stratifier in proxy mode
|
10 years ago |
Con Kolivas
|
099701f26b
|
Set initial start time in the stratum loop to zero to guarantee we get a new workbase immediately
|
10 years ago |
Con Kolivas
|
01240af44d
|
Implement basic get_transactions and new get_txnhashes calls
|
10 years ago |