3638 Commits (11baa82470153c1ad98b075fa75ef4598ca1d590)
 

Author SHA1 Message Date
Con Kolivas eb9c7ffd5a Merge branch 'master' of bitbucket.org:ckolivas/ckpool 8 years ago
Con Kolivas 6587ac2817 Process all hangups after all messages in proxy_recv. 8 years ago
ckolivas 482a8c7d67 Check for existing transactions from the local bitcoind when adding them to our hashlist and submit them only if they're not known about. 8 years ago
ckolivas a772d388f4 Unset wbincomplete sooner to minimise duplicate passes through check_incomplete_wbs 8 years ago
Con Kolivas 3d2c9efcd7 No messages should be postponed any more to increase chance of local block submission and smooth out packet sizes. 8 years ago
Con Kolivas be33c087b9 Propagate workinfo and blocks to all connected nodes as well allowing complex mixed remote server and node infrastructures. 8 years ago
Con Kolivas 2b821bf70e Propagate enough information to allow nodes to be attached to remote servers. 8 years ago
Con Kolivas 137f0b7842 Use a virtualid to reference subclients in remote servers to be able to speak to the upstream pool, allowing them to work with passthroughs. 8 years ago
Con Kolivas 063170f4c9 Reset best shares on accepted remote block submit as well. 8 years ago
Con Kolivas 5ca1781355 Include client id in the key for remote workbases to avoid collisions from multiple remote servers. 8 years ago
Con Kolivas db671316da Age any older workbases rather than just one in add_base by dropping and picking up lock again, and check correct workbase for a readcount. 8 years ago
Con Kolivas eb3f0b2a06 Send block submission data remotely before attempting local submit to not delay propagation with the local verification time. 8 years ago
Con Kolivas 58711dd729 Remove blockhash from submit_block message to generator. 8 years ago
Con Kolivas 9d8c45930b Check inconclusive block submits against their height hash to see if they were already submitted via other means. 8 years ago
Con Kolivas e62c59aade Demote info message in json_rpc_call 8 years ago
Con Kolivas 87aa55cadb Fix memleak with remote workinfos. 8 years ago
Con Kolivas 649d5ff022 Clear_workbase on failed rebuild_txns in proxy mode instead of just freeing the wb memory. 8 years ago
Con Kolivas cd9762d875 Handle the response from generator block submit directly, avoiding the need for a complicated list lookup of block finds. 8 years ago
Con Kolivas fe0549953d Merge branch 'master' of bitbucket.org:ckolivas/ckpool 8 years ago
Con Kolivas d8274d6a1c Allow rebuild_txns to succeed with no transactions listed in the workinfo. 8 years ago
ckolivas f1a0be6994 Use a separate lock for the transaction table. 8 years ago
ckolivas d5f727637c Check for readcount in ageing as well. 8 years ago
ckolivas 10ff10fea0 Use a readcount reference count with workbases to avoid holding the workbase lock and recursive locks. 8 years ago
Con Kolivas d3ab1c5167 Check a txn hasn't already been added to a txntable. 8 years ago
Con Kolivas 53805d7a6d Check and fill incomplete workbases once we receive more transactions. 8 years ago
Con Kolivas 7d8417774f Properly drop subclients when the parent dies by testing whether the parent exists instead of the masked subclient id. 8 years ago
Con Kolivas 3c67dff032 parse_remote_auth was holding a client reference for no good reason preventing remote clients from ever being dropped. 8 years ago
Con Kolivas 3ec08eca6e Don't drop passthroughs as well for not authorising now that their client instance is kept around. 8 years ago
Con Kolivas 22ad7d440a Filter out node.method from going to clients by sending json and intercepting that, avoiding the need to dump it and reencode it. 8 years ago
Con Kolivas 8b7e3b68bc Whitelist clients that have already submitted valid shares in redirector mode by IP and redirect them immediately once they have authorised. 8 years ago
Con Kolivas 556619cfe6 Redirect miners in redirector mode after returning their share results. 8 years ago
Con Kolivas 32499cf34e Failure to get a response to getrawtransaction gives a http error even though it then gives a meaningful response. Drop warning to debug in get_txn. 8 years ago
Con Kolivas 039efd9266 Correctly initialise gbtbase to 0. 8 years ago
Con Kolivas 30e701e829 Count remote users and workers in stats. 8 years ago
Con Kolivas 949b0313b0 Store remote worker and user counts. 8 years ago
Con Kolivas d4fcb0be72 Use a separate mutex for unaccounted shares. 8 years ago
Con Kolivas 20b31f79da Submit transactions to the local bitcoind again when purging them from our transaction table in case a remote workinfo uses them again but our bitcoind has had them removed from the mempool. 8 years ago
Con Kolivas b42fa0a511 Show when transaction building has failed only once in a row and display when rebuilding has succeeded again. 8 years ago
Con Kolivas 499793c94c Add timestamp to throbber. 8 years ago
Con Kolivas 062cb52cbe Abstract out get_timestamp. 8 years ago
Con Kolivas 0a45f8f32c Unnecessary warning. 8 years ago
Con Kolivas 750abcddb6 Properly request transactions we don't find in the local bitcoind and demote the message for failed get_txn to notice only. 8 years ago
Con Kolivas 07d57d683d Prevent passthrough and node etc. failure in generator at startup due to proxy recruit change. 8 years ago
Con Kolivas f40ea70290 Disallow passthroughs on nodes and remote servers as well. 8 years ago
Con Kolivas 90acd45677 Try to find transactions from local bitcoind while rebuilding transactions. Set the refcount for local transactions to 5 only to make remote sending more likely. 8 years ago
Con Kolivas d4869fb5d5 Only send transactions if there are any in the array. 8 years ago
Con Kolivas 3e954e225f Parse upstream requests for transactions, returning any found in the remote's transaction table. 8 years ago
Con Kolivas 68e46be39d Parse remote requests for transactions, returning as many as can be found in the local transaction table. 8 years ago
Con Kolivas 7f19552ee6 When we cannot find a transaction to rebuild_txns, try to get the data from our bitcoind and if it's not available make a transaction request of the upstream or remote pool for it - response yet to be implemented. 8 years ago
Con Kolivas 4cd432475d Decrease the parent's recruit count after sending the stratifier the new subscribe so it can get an accurate headroom count before requesting more proxies. 8 years ago