On Wednesday 09 April 2008 17:16:49 you wrote: > We were discussing this problem yesterday in IRC and its probably > something that will be looked at the hackathon next week. > > Do you have a test case for this by any chance? I don't :(
No ... no test case, though just putting the servers into production does the trick. ;-) We're getting the hangs pretty regularly, so we can probably provide some additional instrumentation if that helps. > This problem seems to occur in both single and multi thread build and I'm > suspecting that the problem could be in the table expansion logic > though I can't say for sure (well, tracing h_next assignments kinda makes > you suspect it). > > I've modified one of our memcached servers to allocate enough buckets > relative to the number of records that we wont exceed so that the expansion > logic wont be triggered (just to see if it'll hang after a week or so > of operation). What about the opposite test ... forcing a server to do the expansion more frequently then normal ... perhaps that will provoke the error? Let me know what I can do to help. thanks, Miguel
