Date: Fri, 25 Sep 2009 19:39:46 +0000
From: Jon Brinkmann
Subject: Re: 2.5m obslog 20090924 (MJD 55099)
On Fri, Sep 25, 2009 at 12:01:01PM +0000, alaina@apo.nmsu.edu wrote:
>
>                        APACHE POINT OBSERVATORY SDSS 2.5M
>
>                                  OBSERVING LOG
>                           Thursday September 24, 2009
>
>                                   (MJD 55099)
...
> host2 crashed, and we had to restart it.
It didn't "crash".  The kernel was having difficulty handling all the
network sockets and was consuming all the CPU time.  Not only was it slow,
but sockets were being closed while still in use.  Rebooting took care
of the problem.
It would also help if the day staff would log out of sdsshost2 at the
end of the day.
...
> DO NOT log into hub25m from sdsshost2.  It will make host2 be overloaded and
> crash.  ssh directly into hub25m.
This is good advice anytime.  You should never "log through" unless it's
absolutely necessary, e.g., to get to a computer  on a private network.
Jon