[gridengine users] qhost load reporting is way off on some Fedora Core 10 nodes, seems fine on FC12, and Centos5.5

Dave Love d.love at liverpool.ac.uk
Sat May 14 12:51:18 UTC 2011


William Deegan <bill at baddogconsulting.com> writes:

> Greetings,
>
>
> Any idea what might cause this?
> host1                  lx-amd64        4    2    4    4  3.04   15.7G    6.0G   37.2G  377.4M
> host2                  lx-amd64        4    2    4    4  1.03   15.7G    4.7G   37.2G  444.4M
> host3                  lx-amd64        4    2    4    4  1.81   15.7G  853.9M   37.2G  227.2M
> host4                  lx-amd64        4    2    4    4  1.02   15.7G    1.9G   37.2G    2.8G
>
> [deegan at nafta4 ~]$ ssh host1 uptime
>  16:56:34 up 21 days,  7:23, 11 users,  load average: 3.13, 3.03, 2.85
> [deegan at nafta4 ~]$ ssh host2 uptime
>  16:56:43 up 156 days, 22:38,  9 users,  load average: 5.00, 5.04, 4.87
> [deegan at nafta4 ~]$ ssh host3 uptime
>  16:56:46 up 150 days,  1:55,  6 users,  load average: 3.01, 3.00, 2.82
> [deegan at nafta4 ~]$ ssh host4 uptime
>  16:57:02 up 156 days, 22:22, 18 users,  load average: 7.48, 7.32, 7.20

I guess you have to debug it, ensuring the data (frpm /proc/loadavg)
aren't stale.  I don't know why it doesn't use getloadavg(3), but it
should give the same.

-- 
Excuse the typping -- I have a broken wrist



More information about the users mailing list