Nigel Kukard wrote: >I'm unable to reproduce anything similar to this on our side, we have >some Policyd boxes in production which handle a very large number of >mails per day and maintain about 500 connections to a single MySQL >server. I've seen maybe one of those errors in Policyd. Can you maybe >shove Policyd into debug mode and see if it appears any query is getting >stuck? possibly may be an idea to tcpdump -w the traffic out and see if >there are any PSH's or repeated packet transmissions which may indicate >packet loss?
I'll get on to that. >The MySQL docs say the most common cause of this error is a timeout... >http://dev.mysql.com/doc/refman/5.0/en/gone-away.html Yes, I'd found that, but am at a loss to figure out why. Of the three mail boxes, two are Xen guests, one is a bare metal machine. I've tried the backend as a Xen guest and running bare metal. I've seen the problem with all permutations of network - xen-xen, xen-real, real-xen, real-real. That to me suggests that it's not a network problem, plus I see no other network issues. I'd upped the connection limit for MySQL, and I'm sure I've not been reaching that. One difference is that policyd is running on the same system as the DB, while postfix is making network connections. I'm thinking it's not policyd that's the problem, just that the extra load policyd creates is unmasking it. -- Simon Hobson Visit http://www.magpiesnestpublishing.co.uk/ for books by acclaimed author Gladys Hobson. Novels - poetry - short stories - ideal as Christmas stocking fillers. Some available as e-books. _______________________________________________ Users mailing list [email protected] http://lists.policyd.org/mailman/listinfo/users
