[ovirt-devel] [ OST Failure Report ] [ ovirt master ] [17.02.17 ] [ external deoendencies missed ]

[Barak Korren]

On 28 February 2017 at 13:52, Michal Skrivanek <michal.skrivanek at redhat.com>
>
> Shouldn’t be a problem to have a full cache of all CentOS packages on the same host or somewhere close for CI. We do not need a different sync for subsequent runs.

We have that, but OST is not only used in the CI system.

> No, I wouldn’t want to remove that. Sure. But if it keeps the cache then why would it take a long time to sync once you do the initial first run?

Well I guess Its a UX issue, do we want a person running OST for the
1st time ever to sit around for a couple of hours to wait for
everything to sync? Do we want to take up this much disk space?

Also the rate of change is CentOS and the other upstreams can be quite
high, so many packages can change between subsequent runs, of which
only a fraction is relevant to oVirt. Some packages also take a heavy
toll bandwidth-wise when they are updated, consider the LibreOffice
family of packages for example...

In any case I don`t disagree that the white list approach has its
issues. But over time we've found that actual failures that arise from
it are few in far between, and are typically easy to resolve.

-- 
Barak Korren
bkorren at redhat.com
RHCE, RHCi, RHV-DevOps Team
https://ifireball.wordpress.com/