Koozali.org: home of the SME Server
Obsolete Releases => SME Server 7.x => Topic started by: wkdr3d on April 22, 2008, 07:55:13 PM
-
Did all the updates to sme on monday morning, everything seemed fine but the server became very unresponsive last night and ive spent many hours looking and reading existing bug reports. I have been able to narrow it down to email (If i cut of smtp at the firewall the server springs back to life) the qsmtpd logs show lots of Too many connections: 40 >= 40. Waiting one second.
i have reduced the amount of allowed smtp connections with the following commands.
config setprop qpsmtpd Instances 5
config setprop qmail ConcurrencyRemote 5
config setprop qmail ConcurrencyLocal 10
signal-event email-update
have also disabled clam and spam assassin and the early_talking plugin in the peers/0 file to try and rule different things out
but it still is slooooooooow and and giving the same 40 >= 40 errors in the logs
The server is a duel xeon 3ghz with 4 gig of ram on SAS discs so spec really shouldnt be an issue in this (And hasnt been untill i did the updates)
My version of qpsmtpd is
rpm -q smeserver-qpsmtpd
smeserver-qpsmtpd-1.2.1-54.el4.sme
ANY help with this would be greatly appreciated , i have a LOT of users on the box and its been rock solid from install.
Regards
-
I think the ones that can help you will be watching the Bugtracker.
-
Ok i have done that, I was unsure whether to post as a bug or not, Not done this before and been using sme since version 6 without ever needing to. Now thats testament to its stability once setup correctly :)
-
I agree with you. I have been on SME so long that I don't remember the first version # I used. In the old days when things weren't patched like today I had a server stay up for over 500 days. I find that the software is written well enough that a lot of forum posts are misconfigured servers, basic network problems, etc. As such I think some of us less knowledgeable can help and maybe save the Dev's from needing to waste time on it.
In the case of this update, no configuration was changed by the user. In this case code changes got us into the problem and it will probably take code changes to get us out. (Some of the code changes need to happen in the contribs.)