Are you sure /etc/profile gets read, especially during boot? I'm not that sure.. Anyway you can verify that the changes took place with:
cat /proc/pidof dovecot/limits|grep files
On Fri, 2009-02-06 at 17:07 +0000, Jake Grimmett wrote:
Hi Mark,
taking your advice, I've put "ulimit -n 8192 > /dev/null 2>&1" into /etc/profile. The problem is horribly intermittent, so it's now just a question of waiting...
Anyhow, many thanks for the tip - I'll post something to the list if the problem comes back !
Jake
On Friday 06 February 2009 15:24, you wrote:
I reported a bug like this on 23rd Dec last year and Timo replied 8/1 this year. However, I've seen this bug very infrequently and I'm not sure that you're hitting it - it's more likely that your ulimit -n is too low - try setting it to around 8192 (defaults to 1024 I think). It should be around 3* max conns you expect to be on the safe side of things.
Mark
-- Mark Zealey -- Shared Hosting Team Leader Product Development * Webfusion 123-reg.co.uk, webfusion.co.uk, donhost.co.uk, supanames.co.uk
This mail is subject to http://www.gxn.net/disclaimer
-----Original Message----- From: dovecot-bounces+mark.zealey=webfusion.com@dovecot.org [mailto:dovecot-bounces+mark.zealey=webfusion.com@dovecot.org] On Behalf Of Jake Grimmett Sent: 06 February 2009 15:15 To: dovecot@dovecot.org Subject: [Dovecot] help: "dovecot: pipe() failed: Too many open files"
Dear All,
I'm having an intermittent problem, with dovecot slowing to a crawl. Restarting dovecot fixes the problem.
Looking at the maillog, I see loads of these errors....
dovecot: pipe() failed: Too many open files dovecot: Temporary failure in creating login processes, slowing down for now dovecot: Created login processes successfully, unstalling
looking at the processes, I can see ~200 imap-login's and about ~700 imap's there are 1000 email accounts on the server - half of which are in regular use.
The server is Dell PE1950 running dovecot (1.0.7-2.el5) as supplied with Centos 5.2 (~Redhat AS 5.2) in 64 bit mode. I've upgraded the server from 4 to 8GB of RAM, and from a dual core 1.5Ghz xeon, to a quad core 1.5GHz xeon, and this does not seem to help.
The following settings were increased by me to try and stop this problem - but did not have an obvious effect:
login_process_size = 128 login_process_per_connection = yes login_max_processes_count = 512
Other parameters were left commented out:
#max_mail_processes = 1024 #auth_process_size = 256
If anyone has any ideas about where I might be going wrong, please let me know..!
many thanks
Jake
-- Dr Jake Grimmett Head Of Scientific Computing MRC Laboratory of Molecular Biology Hills Road, Cambridge, CB2 0QH, UK. Phone 01223 402219 Mobile 0776 9886539