Quoting Ian B porjo38@yahoo.com.au:
--- On Fri, 14/1/11, Tom Hendrikx tom@whyscream.net wrote:
I had this problem too when I setup a VMware guest last month (noticed through strange spikes in munin's ntp monitoring, not through dovecot logs). After some research I found out that I had VMware time synchronisation (host to guest) enabled, and NTP in the guest running. The ESX host clock was some 90+ seconds behind, so VMware stalled my clock once in a while, after which NTP corrected it again.
I asked the ESX admin to fix the clock, but disabled the VMware synchronisation and now use only ntp in the guest. Never looked back.
So in stead of 'ticking all the boxes', explain your setup. You have NTP enabled, but what does VMware do?
Thanks for the post. I've checked my setup - the guest is not set to
sync with the VMware host. NTP is running with local clock server
commented out & 'tinker panic 0' set.It sounds like this is not a Dovecot issue so I'll stop posting to
this list now and seek help elsewhere.What negative effects to Dovecot would I expect to see where the
system clock is constantly changing?
my experience with CentOS/Dovecot was that when the clock is reset
back, Dovecot writes a message into /var/log/messages saying, IIRC,
"Time just went backwards. This could cause problems so I'll just kill
myself now." And mail transport stops.
A horrid nuisance to debug. I don't know about it changing in the
other direction.
Dave
-- "It is no measure of health to be well adjusted to a profoundly sick society." Krishnamurti