On Mon, 21 Apr 2008, Vinny Wadding wrote: > Hi guys, > I have just installed XMail 1.25 and am not seeing some odd behaviour > from it. It has taken me a while to try and track down what is > happening, but here it is from what I can find. > > The new server is put live and works extremely well. After several > hours, I get alerts saying that the system memory is exhausted. The > memory footprint of XMail stays consistent at about 85mb - the rest of > the system memory is all allocated to Buffers and Cache. There is > virtually no swap space used, but a higher than expected load average on > the system. > > I have taken the server offline so there is no traffic running to it. > On a reboot the memory is released, but as soon XMail is restarted the > Cache and Buffers climb back up and use all the memory again. That > would eliminate an external influence on the server, and suggest > something that is already on the XMail queue. > > I wiped all frozen spool from the server and restarted again. Same as > above, the server released the memory but as soon as XMail started it > started to grab all free memory for Cache and Buffers. This would > suggest something that XMail is still trying to process? > > I trawled the spool queue and found several mails that appeared to be > "stuck". It would appear that XMail would repeatedly try and process > these, it would not as I could see spool files being created and removed > as it goes thorough the motions. I have downloaded the spool queues > and had a look through them offline, it would appear that the mails that > are causing an issue on this server are coming from badly set up > domains. I ran these domains through dnsstuff to see how they were set up. > > One of the domains was set up with no MX record and no A records. Even > when manually submitting test mail for this domain, it would be accepted > and then causes XMail to overreact. > The rest of the domains I saw on the spool queue had minor anomalies, > but when submitting manual mails via telnet for these domains it did not > seem to cause the same reaction as the domain with no zone information. > > Obviously wiping the spool queue and rebuilding it from the source > example resolves the issue and the behaviour returns to normal. > > Is this a known behaviour or "feature" in 1.25? Has anyone else seen > this happening? Is there any way of being able to stop this behaviour?
I don't think so. The "eat all the server's memory" feature will not come out till 1.26 :) Which OS is that? Why don't you post one of your telnet transactions that are creating problems? Talking about "some domains" and "some emails" is a bit vague, since it does not allow anyone to replicate your box results. This is my XMail BTW: VSZ RSS 18676 5184 - Davide - To unsubscribe from this list: send the line "unsubscribe xmail" in the body of a message to [EMAIL PROTECTED] For general help: send the line "help" in the body of a message to [EMAIL PROTECTED]