[prev in list] [next in list] [prev in thread] [next in thread] 

List:       npaci-rocks-discussion
Subject:    [Rocks-Discuss]  CentOS mismatch on frontend and nodes
From:       Ari Berman <ari () bermanism ! com>
Date:       2015-01-24 18:39:58
Message-ID: CANFi+-Ye1LcUbSCb1mqjaQHaw1Eo5De9ghLnk+yXitR2trjz7Q () mail ! gmail ! com
[Download RAW message or body]

Hi Fellow HPC experts,

I'm hoping someone has an easy answer to my conundrum. I updated my
frontend some time ago (Rocks 6.1.1) using yum update with base and updates
enabled. It updated to CentOS 6.6, which isn't a problem. I got busy, time
passed, and someone else ran yum clean in the mean time, clearing out the
rpm cache from yum, which I was meaning to put into /export/rocks/install
for inclusion into the node installation. Now, it is all gone, and CentOS
is at two different versions on the front end and the compute nodes, along
with two very different versions of the kernel. This is bad for things like
ZFS, MLNX-OFED, and cuda, which I'm using across the cluster.

Does anyone have a brilliant idea for how to re-download the rpms so that I
can get all the nodes to the same version level as the front end, without
reinstalling the front end? No, I didn't make a backup roll like I should
have. :/

Thanks in advance!

--Ari

-- 
Ari E. Berman, Ph.D.
General Manager of Government Services and Principal Investigator
BioTeam, Inc (http://www.bioteam.net)
Toll-Free: 877-246-2992 x110
Direct: 512-366-8194
email: ari@bioteam.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.sdsc.edu/pipermail/npaci-rocks-discussion/attachments/20150124/0deb0698/attachment.html \



[prev in list] [next in list] [prev in thread] [next in thread]