On Fri, 2010-03-26 at 18:44 +0100, Ivan De Masi wrote:
Am 26.03.2010 18:37, schrieb Vojtěch Zeisek:
Dne Pá 26. března 2010 18:28:37 Ivan De Masi napsal(a):
Hello,
I am wondering since some time how others cope with patch management on OpenSUSE Clients, when you have a bunch of PCs in a network. Somebody who can recommend a open source tool to cover the basics (e.g. a tool automating updates with zypper over ssh or something similar)?
Hello, in YaST | Software You can set automatic upgrade in certain time. And it works. :-) Another possibility is for example to use Konsole, which can send one command to multiple machines. You can then send "zypper dup" to all Your computers. I hope it helps little bit. :-) Best regards, Vojtěch Zeisek
Hi Vojtěch,
thanks, I know this :-)
What I'm looking for is a piece of software to do "zypper up" for example on 15, 20, or 2000 machines at the same time in a controlled way (building groups of machines before that or so). Now I have to log in via ssh on every machine and do my
# zypper lp # zypper lu # zypper up
an so on...
Shure - if my scripting know how would be better I would write my own mechanism for automating these steps an add some more...
If you want to run a program either simultaneously or sequentually, have a look at tentacle...
However, considering doing upgrades, i'm not shure if i would recommend doing it unattended and automagically
Most problems i encounter (having 200++ machines) come to light after a mandatory system reboot. Not all systems come up. Sometimes they just fail to die (nasty habbit of md) or there's something wrong with one of the kernel-modules.
Bad enough if just one machine failes, but in my case most of the hardware is running XEN-hypervisors, so if they fail, all of the clients-machines are lost as well.
And if something goes sour with apache or php (ften some undocumented change in kerberos-code), i rather find out with the first machine, and put the pathes for all other machines on hold, while repairing the first one.
Furthermore if the machine with either the nfs-share with the repo-data or the local httpp-install-server goes down while an upgrade is going on, on another machine.... Well it wouldn't speed up the process ;-)
That's why i do those upgrade on the last friday of th month: most people leave early on that day...
hw