How do _you_ handle applying patches to lots of machines?
2007-12-25 11:28:00
highlights. There are two basic solutions -- most people seem to just
relay on rsh/rdist scripts and hosts.equiv.. others use a program called
dmake which is available from ftp.scri.fsu.edu, dmake appears to be an
"enhanced" rdist.
The essential thing was to put the patches on n exported NFS volume. Saves
loads of time. This time around I have to do patches so piecemeal that I
will likely just go to the individual machines. Professors can be hard to
schedule. :)
billt@dat.com (Bill Townsley) is an ECKERD graduate!
and commented that doing an rsh target_machine pkgadd.. with an NFS volume
would work.
wbmoore@fedex.com (Walter Moore) said:
>Walter here :)
>
> put the recommended patches in a directory that I export to all the
>client machines (I call it /opt/patches/2.4_Recommended). Then I add
>in the other patches I want each of my clients to have to that directory.
>when I append the id's of the extra patches to the end of the file
>patch_order. Then in a script I rsh to each client and run install_cluster
>in the directory in question.
>
>
>#!/bin/sh
>for i in `nisgrep addr='161.135.5[12]*' hosts.org_dir | cut -f2 -d ' ' |
grep -v router` ; do
> echo $i
> rsh $i "cd /opt/patches/2.4_Recommended; ./install_cluster"
>done
{what ARE the chances of there being two Walter Moores on this list?}
rloftin@engsys.mc.xerox.com (Ron Loftin) suggests:
>My approach is to maintain a directory of patches that is available via
NFS
>and automounter with patches, and have a script running in cron nightly
that
>scans the directory for new patches, sends e-mail to users of the
machine(s)
>that need patches, and creates "at" jobs to install the patches. This is
a
>bit time-consuming, but allows users to log off and save their work,
rather
>than go through reboots with applications left up.
tkevans@eplrx7.es.dupont.com (Tim Evans) suggested dmake, which acts as a
distributed make. Looks pretty neat.
My apologies to anyone I missed.. I'm still reading suggestions...
This is a partial list of those who sent me replies. Thank you one and
all!
+--
From: tkevans@eplrx7.es.dupont.com (Tim Evans)
From: popp@luey.redars.ca.boeing.com (Jeff Popp
From: billt@dat.com (Bill Townsley)
From: Laura Taylor <ltaylor@voom.bbn.com>
From: "Randall S. Winchester" <rsw@Glue.umd.edu>
From: Rich Kulawiec <rsk@itw.com>
From: "Henry Katz" <hkatz@lehman.com>
From: Mark Fogelson <fogelson@sentientnet.com>
From: Shaun Welch <swelch@home.net>
From: Peter Gray <pdg@uow.edu.au>
From: seanm@sybase.com (Sean McInerney)
From: jackg@calfp.com (Jack Goldsmith)
From: rloftin@engsys.mc.xerox.com (Ron Loftin)
From: wbmoore@fedex.com (Walter Moore)
From: David.Stewart@ee.ed.ac.uk
+-------------------------------------------------------------------
Walter R. Moore - Assistant System Administrator
http://www.eckerd.edu/~moorewr
Comments
Got something to say?
You must be logged in to post a comment.
