[Gluster-infra] Added yum-cron rpm to the Rackspace slaves

[Justin Clift]

On 30/08/2014, at 10:43 AM, Michael Scherer wrote:
> Le samedi 30 août 2014 à 00:40 +0100, Justin Clift a écrit :
>> Just a small FYI.  The Rackspace slave VM's (slave2*) now have
>> yum-cron installed and enabled.  In theory this should mean the
>> packages on them are automatically kept up-to-date each night.
>> 
>> Doesn't seem like it reboots things whenever a new kernel is
>> installed, but it's a step in the right direction anyway. :)
> 
> I think we still need to take a look at the server and reboot them from
> time to time, for new kernel update, and because it is a easier
> procedure to reload library. 
> 
> But it is a bit hard to decide if you need a reboot or not if you do not
> see the update being applied :/
> 
> ( even if rpm -qa --last come handy for that ).

There is another approach that might work too.  There's a yum
package called "yum-updateonboot":

  $ yum info yum-updateonboot
  Loaded plugins: changelog
  Available Packages
  Name        : yum-updateonboot
  Arch        : noarch
  Version     : 1.1.30
  Release     : 17.el6_5
  Size        : 21 k
  Repo        : updates
  Summary     : Run yum update on system boot
  URL         : http://yum.baseurl.org/download/yum-utils/
  License     : GPLv2+
  Description : Runs yum update on system boot. This allows machines that have been turned
              : off for an extended amount of time to become secure immediately, instead of
              : waiting until the next early morning cron job.

We could install that + put a reboot in cron.  eg every Saturday
at midnight

That'd probably fix the problem.

>> Side note - Some of the regression tests failed tonight with
>> a weird error about being unable to clean the workspace.  I've
>> logged into those and rm -rf /home/jenkins/root/*, which should
>> fix the problem.  Feel free to do that again if the error
>> crops up in the future. :)
> 
> Doesn't that mean that
> 1) root did create a file
> 2) jenkins manage to remove the permission to delete file 
> 
> and in both case, we should investigate a bit more ?


Well yeah, we could do that too.  Next time then. :)

+ Justin

--
GlusterFS - http://www.gluster.org

An open source, distributed file system scaling to several
petabytes, and handling thousands of clients.

My personal twitter: twitter.com/realjustinclift