On Thu, Sep 13, 2012 at 11:13 AM, Tony Graziano <
[email protected]> wrote:
> We've seen a small number of instances with 4.6 where sending the server
> profiles seemed to resolve the issue. This was almost always after an
> update to the sipx system is made.
>
> So I need to ask, is it justifiable to add a watch mechanism that says
> "new package installed/send server profiles!" as an automatic function?
> Does this break anything else?
>
Good suggestion, here's my input we can create a jira on
Details:
============
Technically "Send Profiles" from UI does 3 things:
1.) Regenerate configuration meta data
2.) Run cfengine on appropriate systems
3.) Resync MongoDB with postgres
If you restart sipXconfig it actually performs #1 and #2
Proposal:
============
1.) If a package updates sipxconfig plugin it should restart sipxconfig
which runs step #2 and #3
Command to be added to rpm spec post update
service sipxconfig condrestart
2.) If a package does not already do #1, but updates cfengine scripts it
should run step #2 on the system it updates
Command to be added to rpm spec post update
sipxagent
3.) If a package does not already to #1 or #3, it should should restart
appropriate services to load new code.
Command to be added to rpm spec post update
service sipxfoo condrestart
Requirements:
* If multiple packages are updates within a given time window, we should
make sure we do not run these commands more than once.
Summary
============
With all this implemented, from an administrators perspective, system
should be able to handle yum updates w/o additional commands/intervention
but they should expected appropriate service disruption.
_______________________________________________
sipx-dev mailing list
[email protected]
List Archive: http://list.sipfoundry.org/archive/sipx-dev/
