[Bi-users] bi1 login node rebooted earlier today

[Kent Engström]

Dear Bi users, for your information,

the bi1 login node had a high load during the morning hours and felt
quite sluggish at some times, but we found no particular process to
single out for attention.

At around 12:20, it crashed and rebooted, and seemed to be OK after
that.

Even if this might not be the cause this time, we would like to remind
you to allocate compute nodes (using sbatch or interactive as it suits
you) when you run CPU- or I/O-intensive processes for more than a few
minutes. Doing so helps both your own perfomance, and the performance on
the login node (for the things that need to be done there).

Those of you who are doing rsync:s from the old filesystems about to be
retired to the new filesystems, should do that on compute nodes.


Best Regards,

-- 
Kent Engström, National Supercomputer Centre
kent at nsc.liu.se, +46 13 28 4444