[torqueusers] RPM's for torque

Nico Kadel-Garcia nkadel at merl.com
Wed Sep 29 14:42:28 MDT 2004


Cool, I wasn't aware of "BioRPMS".

If it's OK to have a MOM on the server, cool. It doesn't match our
environment, but I got pulled into maintaining this as some of my time got
allocated to supporting a pair of active clusters. (The previous maintainer
was consulting and has fled for his home country.) As an old sys-geek, I've
overly familiar with building basic RPM's, but didn't have time or desire to
do one from scratch, so I worked from the old OpenPBS SRPM published by
SuSE. That had some..... interesting SuSE behavior in it.

The "$1" handling is a good idea, I'll integrate that.

I've also been looking at the node/serverdb handling for "pbs_server -t
create" in the init scripts. I'd like to handle that better Currently, if
there is a "$pbs_home/server_priv/nodes" file and no
"$pbs_home/server_prive/serverdb" file, the serverdb is created and
everything is hunky dory, but pbs_server goes on running with options that
are not normal to its command line. If both files happen to exist, the
"pbs_server -t create" option erases *both* of them. This.... is kind of
ugly, and the error messages given if the serverdb doesn't exist are kind of
unhelpful.

I'm not familiar with BioRPMS: Ghods know I've had to deal with
Bioinformatics from a systems level, both as a systems geek doing medical
research and listening to my wife's laments when she did bioinformatics for
genetic sequencing. BioRPMS seems like a reasonable repository. Is there
someone over there who could take over the .torque package bundling if I
pass along hints?

----- Original Message ----- 
From: "Bent Terp" <Bent.Terp at biosci.ki.se>
To: <torqueusers at supercluster.org>
Sent: Wednesday, September 29, 2004 1:39 AM
Subject: Re: [torqueusers] RPM's for torqueOn Tuesday 28 September 2004
02:33, Nico Kadel-Garcia wrote:
> Hi, folks! I've been trying to bundle up Torque for RedHat use in our lab,
> and have built up RPM's and SRPM's. They're not perfect, but they're
> working alright for me.

Hi Nico! Seems not too unlike what I've had in BioRPMS for a while - except
that I really should do an update to the latest version.

Apart from three small issues, it actually looks a bit better than my spec
file.

1) There's nothing inherently wrong with having a mom on the server, indeed
that's quite normal behaviour for small clusters I'd say.

2) Your scripts seem to ignore $1 and thus remain ignorant of whether an
install, upgrade, or remove takes place.

> review them. I haven't put PGP signatures on them yet, but I've in the
> midst of getting buy-in to set up repository for those.

Would you be interested in joining BioRPMS instead of setting up your own
repository?

-- 
MVH / With kind regards,
        Bent Terp <Bent.Terp at biosci.ki.se>
        System Administrator
        Bioinformatics and Expression Analysis Core Facility
        Karolinska Institute, Department of Biosciences
        Hälsovagen 7-9, 141 57 Huddinge, Sweden



More information about the torqueusers mailing list