Published on: 2011-09-09T20:00:30+00:00
On September 9, 2011, Luke-Jr posted on a forum about an issue he was facing with two bitcoinds running on the same system on Eligius. One of these bitcoinds is functioning as a hub node, relaying network activity between hundreds of nodes that Eligius peers with. The other bitcoind is a work node, responsible for managing mining operations and only connecting to the hub node.The problem at hand is that the hub node has been experiencing unexpected crashes recently, and the cause is yet to be determined. Strangely, there haven't been any recent changes made, making it difficult to pinpoint the root cause of this issue. In response to Luke-Jr's post, another user humorously requested a stack trace as proof.When the hub node crashes, the work node loses its connection as well. This leads to the work node responding with "not connected" messages when queried for getworks. What's interesting is that the sockets used for these messages are not properly cleaned up, resulting in an overflow of the system's socket limits. Luke-Jr mentions that he is using a heavily patched version, specifically version 0.3.23. It is possible that one of these patches, such as the threaded JSON-RPC, could be causing the bug. However, he hasn't had the chance to reproduce the issue on an unpatched version yet. He does mention that he wanted to bring this up before the release of version 0.4 final, but due to being extremely busy, he may not have the opportunity to do so.
Updated on: 2023-08-01T02:25:38.150301+00:00