Am 21.03.2014 21:46, schrieb Paul J Stevens: > On 21-03-14 21:01, Reindl Harald wrote: >> looks like the random timeouts are really away now >> see sub-thread of "Iphone HTML emails" a few days ago >> >> finally we will see the result next week since these got more after longer >> uptime, however 2 days without any error looks new, not a single connection >> error from the dovecot-proxy after the update > > Nice. Very good to hear. If you can try to profile memory usage, using > rrdtool or similar tools. I'm very curious how memory footprints hold up > in usage patterns that differ from the ones I see myself, read: much > higher concurrencies and more diverse clients
can you elaborate rrdtool in context of memory usage? i am happily monitor that on our production servers, currently only aware of it in context mailgraph / smokeping, systemd runs anything fine as service (some of mine are "while (1==1) loops" in php-scripts with autorestart in case of fatal errors so anything you have for monitoring -> throw it to me _________________________________________ for what i have seen now below 200 MB for imapd and way lower for the other services, the stats are starting now from zero because update of nss and processes listed in "lsof | grep DEL" restarted but as far as i remember 3.1.13 has the best values i have ever seen since dbmail-3.0 honestly i think even dbmail-2.x was affected by that timeouts because i remember client complaints of MUA's asking for their passwords from time to time, iexplained that with connection troubles on the client side years ago, not proveable but maybe also cleaned up that rare cases thanks so much for the hard work on that great software, we both know way too much how painful debugging is - i have never seen any maintainer finding reported issues that often by look at the code and nothing else!
signature.asc
Description: OpenPGP digital signature
_______________________________________________ DBmail mailing list [email protected] http://mailman.fastxs.nl/cgi-bin/mailman/listinfo/dbmail
