Hi All.
I had today a problem with my mail server (2.6.18-274.12.1.el5 #1 SMP Tue Nov 29 13:37:35 EST 2011 i686 i686 i386 GNU/Linux, CentOS release 5.7 (Final)). On my Cacti graphs I see that there has been much I/O write on the disks and then there is no more info. Also in logs (messages, dmesg, netconsole) there is no info for about 15 minutes when there was a problem with the server.
I would like to have some info in such situations. Do you know any solutions?
Best regards, Rafal.
Rafał Radecki wrote on 04/12/2012 03:07 AM:
Hi All.
I had today a problem with my mail server (2.6.18-274.12.1.el5 #1 SMP Tue Nov 29 13:37:35 EST 2011 i686 i686 i386 GNU/Linux, CentOS release 5.7 (Final)). On my Cacti graphs I see that there has been much I/O write on the disks and then there is no more info. Also in logs (messages, dmesg, netconsole) there is no info for about 15 minutes when there was a problem with the server.
I would like to have some info in such situations. Do you know any solutions?
Your description is unclear. Did the system recover, or did it have to be power cycled or otherwise rebooted?
A hang with no log information can sometimes be caused by driver issues.
The first thing to try is a "yum update" to the current and supported version 5.8.
Phil
Phil Schaffner wrote:
Rafał Radecki wrote on 04/12/2012 03:07 AM:
Hi All.
I had today a problem with my mail server (2.6.18-274.12.1.el5 #1 SMP Tue Nov 29 13:37:35 EST 2011 i686 i686 i386 GNU/Linux, CentOS release 5.7 (Final)). On my Cacti graphs I see that there has been much I/O write on the disks and then there is no more info. Also in logs (messages, dmesg, netconsole) there is no info for about 15 minutes when there was a problem with the server.
I would like to have some info in such situations. Do you know any solutions?
Your description is unclear. Did the system recover, or did it have to be power cycled or otherwise rebooted?
A hang with no log information can sometimes be caused by driver issues.
<snip> Or by a failing disk.
Time to use smartctl, if not fsck -c.
mark
On Thursday, April 12, 2012 10:13:00 AM m.roth@5-cent.us wrote:
Time to use smartctl, if not fsck -c.
Another useful fsck variation is fsck -cc which will do a nondestructive 'read-write' to trigger on-drive bad sector remapping. Takes a long time to do, but does the work.
It's documented in the man page under the '-c' option's paragraph, last sentence.
The system was not rebooted, it just was not responsive (ssh) and has a gap in logfiles for 15 minutes. After 15 minutes it started responding.
Best regards, R.
W dniu 12 kwietnia 2012 16:08 użytkownik Phil Schaffner < Philip.R.Schaffner@nasa.gov> napisał:
Rafał Radecki wrote on 04/12/2012 03:07 AM:
Hi All.
I had today a problem with my mail server (2.6.18-274.12.1.el5 #1 SMP Tue Nov 29 13:37:35 EST 2011 i686 i686 i386 GNU/Linux, CentOS release 5.7 (Final)). On my Cacti graphs I see that there has been much I/O write on the disks and then there is no more info. Also in logs (messages, dmesg, netconsole) there is no info for about 15 minutes when there was a
problem
with the server.
I would like to have some info in such situations. Do you know any solutions?
Your description is unclear. Did the system recover, or did it have to be power cycled or otherwise rebooted?
A hang with no log information can sometimes be caused by driver issues.
The first thing to try is a "yum update" to the current and supported version 5.8.
Phil
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
From: Rafał Radecki radecki.rafal@gmail.com
T he system was not rebooted, it just was not responsive (ssh) and has a gap in logfiles for 15 minutes. After 15 minutes it started responding.
Maybe also check the bios/ipmi logs...
JD
-
John Doe -
Lamar Owen -
m.roth＠5-cent.us -
Phil Schaffner -
Rafał Radecki