On Sun, Oct 15, 2017 at 10:15 AM, Barak Korren <bkorren@redhat.com> wrote:


On 13 October 2017 at 14:31, Sandro Bonazzola <sbonazzo@redhat.com> wrote:

2017-10-13 11:29 GMT+02:00 Dafna Ron <dron@redhat.com>:
Adding Eyal, Barak and Sandro and removing Victor.

Personally, I do not mind taking on this task, but I think that I would need help in creating such a task.

I think we are already mirroring everything and refreshing the mirror every few hours.
Issue looks like we are not using them in some jobs.


I did not look deeply into this particular issue, but just to get everyone on the same page.

* We have mirror system that is getting synced every 8 hours and has no known issues
  ATM
* All repo issues you're seeing in OST are due to one of the following two reasons:
  1. We are white-listing packages into the OST environment and the list needs to
     be maintained as package dependencies change
  2. The OST VMs are not blocked from using the upstream CentOS repos/mirrors. And
     the upstream repos are not updated in an atomic fashion

We have ongoing work [1] to fix issue #2 above, it takes time because it requires meticulous work to get all the required things into the whitelist.

Since I send here and there patches to do this meticulous work, I know it's not such a big deal. 
Yes, it's annoying and I have not yet come up with an automated way to do it (I'm sure there is!), but it takes few hours and we can do it once every 2 weeks or so.
It also has the nice benefit of reducing run time, sometimes dramatically.
Y.
 

BTW when you see these issues in OST that are doe to upstream CentOS repos not being updated atomically, it usually correlates with a similar failure in the mirror sync job.

[1]: https://ovirt-jira.atlassian.net/browse/OVIRT-1280
 
--
Barak Korren
RHV DevOps team , RHCE, RHCi
Red Hat EMEA
redhat.com | TRIED. TESTED. TRUSTED. | redhat.com/trusted