Paul J Stevens wrote: > Peter Rabbitson wrote: >> Let me restate my question from the last mail - am I understanding >> correctly that this is not considered a major issue anymore? > > No. Any real leakage that is not a flatliner is a major issue. In your > case there is leakage that scales with the number of messages handled. > DBMail must deal with this or perish. >
Good to hear :) >> Yes the >> test test case might be harsh, but every _morning_ the place where I >> wanted to deploy dbmail moves an average of 2000 messages between 2 >> boxes. In this situation the risk of memory issues is too high for me. I >> have no problem with helping you guys track this further, but unless >> there is an agreement that the behavior I am experiencing is plain >> wrong, I probably have to look for other alternatives to dbmail. > > All your logs pointpoint the same problem point: the core-server code. > > But before you run another test: use G_SLICE=always-malloc. Oh my god... Before you go any further let me rerun the tests. I realized that I have been running with G_SLICE=always_malloc instead of always-malloc. I am terribly sorry, too many years of perl got in the way. _______________________________________________ Dbmail-dev mailing list Dbmail-dev@dbmail.org http://twister.fastxs.net/mailman/listinfo/dbmail-dev
