Hello,
I am using vserver approx. 2 years for hosting almost all my services
(web, mail, application services)
A half year ago I changed my main servers to AMD Athlon 64 Processors
(3700+). With the host operating system I stayed at 32-bit. I always
used lvm2 with reiserfs and with the change I also switched to drbd to
mirror the lvm-partitions to a slave servers for the important
vservers running.
Since the switch to this configuration the stability is not optimal.
Troubles occur in average 1 time a month. Which is not OK, but I
accepted it. But yesterday and the day before yesterday during night
there were troubles on 2 machines within 10 minutes.
What happens? The procedure is allways the same. I monitor the loadavg
of all vservers. Here is a table of the loadavg for the host and 1
guest.
seconds loadavg_guest loadavg_host
0 0.8 0.25
30 3.3 10.77
60 7.5 20.6
90 11.8 30.4
120 15.7 39
150 16.6 41
... (increase about 2 in 30 seconds)
The other guest on the machines are not affected and worked well. The
affected guest had high timeouts with http-connections.
I could not enter the affected guest with vserver ... enter. vserver
... stop did not work. vkill did also not work. (I always had to
reboot per hardware-reset.)
The troubles were always during night hours (with 2 exceptions) where
the load on the servers is minimal.
Here are the program versions I use
Linux version 2.6.17.13-vs2.0.2.1 (gcc-Version 3.3.5 (Debian 1:3.3.5-13))
Debian sarge on the host. Most guest are also sarge, some are gentoo,
some are ubuntu 6.06
drbd-0.7.21
util-vserver-0.30.210
kernel / drbd and util-vserver is self-compiled, all other packages
(incl. lvm2) are from debian sarge main
Does anyone have an idea, what's problem? I don't want to wake up
every night per sms.
Walter
_______________________________________________
Vserver mailing list
Vserver@list.linux-vserver.org
http://list.linux-vserver.org/mailman/listinfo/vserver
Received on Wed Nov 29 12:51:01 2006