launchpad-dev team mailing list archive

[Danilo Šegan <danilo@xxxxxxxxxxxxx>]

Hi Stuart, Robert,

As I mentioned earlier, DBLoopTuner considers lag to be huge when at
least one replication slave is being rebuilt. This makes the script
(usually a migration script, but we also want to move more of our
regular scripts to DBLoopTuner) stall whenever a slave is being rebuilt,
which sucks.

I've filed this as 
  https://bugs.edge.launchpad.net/launchpad-foundations/+bug/622670

I've also realized that we do have a case where this is harming us as
well in the first script we kind of moved to DBLoopTuner:
rosetta-pofile-stats-daily.py.

https://bugs.edge.launchpad.net/rosetta/+bug/622668

I am not sure how much of it is due to slowness, and how much is it due
to DBLoopTuner acting up, but it's definitely hurting us.  Also, I am
not sure if the DB sleep() time is also taken into account when
determining the next loop size (but it could very well be a bug in
pofile-daily-stats script, so if you don't know the answer of the top of
your head, let's not worry about it yet).

Cheers,
Danilo