[Development] Moving to Gerrit 2.16.9

Kari Oikarinen kari.oikarinen at qt.io
Tue Jul 2 14:43:06 CEST 2019

On 2.7.2019 15.13, Kari Oikarinen wrote:
> On 2.7.2019 14.59, Cristian Adam wrote:
>>> -----Original Message-----
>>> From: Development <development-bounces at qt-project.org> On Behalf Of
>>> Frederik Gladhorn
>>> Sent: Thursday, 27 June 2019 16:11
>>> To: Qt Project Development Mailing-List <development at qt-project.org>
>>> Subject: [Development] Moving to Gerrit 2.16.9
>>> Hi,
>>> Just to keep the ball rolling, we prepared the upgrade to move from Gerrit
>>> 2.16.7 to 2.16.9. I don't expect any real changes, but it's a good exercise for us
>>> to stay up to date and see if the scripting of the upgrade works.
>>> The only real challenge was upgrading Bazel, since every Gerrit version
>>> seems to only compile with one exact Bazel version (roughly).
>>> So far it seems to work nicely for me, running the script which stops gerrit,
>>> pushes the new release and restarts it takes around two minutes, so there
>>> won't be any significant downtime this time around.
>>> Assuming there are no big concerns I'll just do the upgrade tomorrow, on the
>>> test instance it works without problems.
>>> On a related note, now that things are generally working with the new Gerrit,
>>> I was wondering if we want to consider plugins. There is one to add
>>> reviewers based on git blame https://gerrit-
>>> review.googlesource.com/admin/repos/plugins/reviewers-by-blame
>>> and I'll give Gravatar a spin:
>>> https://gerrit-review.googlesource.com/admin/repos/plugins/avatars-
>>> gravatar
>>> We should also consider the various webhooks plugins. Comments
>>> appreciated.
>>> Cheers,
>>> Frederik
>> Hi,
>> I'm getting from Qt Sanity Bot for https://codereview.qt-project.org/c/qt-creator/qt-creator/+/266791
>> The following reply every 5 minutes:
>> **** Worker 'sanity' produced an unreasonable amount of output. You should ask the bot maintainers for advice.
>> The current behavior is a bit loopy, can we do something about it?
> I guess you've managed to get sanity into a situation where it doesn't give a
> sanity score at all.
> Then once the bot looks at the commits it hasn't scored yet, it will repeat that
> again and again.

This isn't actually sufficient to cause the issue, the bot remembers what
commits it has looked at during current run and doesn't look at them again.

So the new attempts only happen when sanity bot is started again and it looks at
all open changes it hasn't reviewed yet.

But looks like for some reason sanity bot's connection (with which it receives
notifications about new patchsets) drops every 5 minutes and causes the bot to
quit. When it is automatically restarted, it then tries to review again because
of missing review.

I manually put a negative review in place, so at least the spam should be over
for this specific commit for now.

> Could you run sanity bot locally and see what is happening to it? There is a git
> hook in qtrepotools to do it when you commit.


More information about the Development mailing list