On Thu, Oct 17, 2019 at 10:08:30AM +0200, Pierre-Yves Chibon wrote:
...snip...
> 2) we need nagios monitoring those certs, or we need to just
tear
> down that cluster if we aren't going to use it (which we are currently
> not).
>
> 3) We could also 'unrepospanner' that repo since we aren't using it
> and put the old one back.
This may be wise, especially considering that I may not have fixed everything
(see the end of this email).
I'm not entirely sure how to do this. Just unset repospanner in the
pagure settings?
> 4) pagure perhaps should gracefully print 'sorry, the repo
is not
> available right now due to a repospanner problem' but otherwise work?
+1 for this, I'm not sure of the size of the work in there but worth looking
into.
Shall I file a pagure ticket? Or you want to?
Also: Patrick said that the cert needs to be upgraded in other places (nodes) as
well, I do not know if running the repospanner playbook fixed it or not though,
so we may still have something broken.
I have received emails from pagure yesterday with:
"""
...
PagurePushDenied: Remote hook declined the push: Performing pre-check...
...
ERR Error syncing object out to enough nodes
"""
Which make me think we are still missing some fix, but I don't know which :(
There's 3 nodes defined in this cluster:
pagure01
batcave01
repospanner-phx2
So, I think we need to run the playbook on batcave and repospanner-phx2.
kevin