As an initial step to winding down TBPL, let's disable log prefetching. This will mean that when import-buildbot-data.py runs on the 5 min cron, we'll just import build-4hr.js, rather than doing that and then also fetching every single failed log from FTP, parsing the log for errors & then generating bug suggestions for each error. If someone selects the job in the TBPL UI, the on-demand log fetching/parsing will still kick in however, so people can still view the error excerpt. This change will: * Reduce the load on ftp.m.o. * Reduce TBPL's daily DB consumption quite considerably. * Reduce the load on the shared cluster on which TBPL resides. * Reduce the load on Bugzilla's REST API, since the bugscache implementation of TBPL is suboptimal, causing us to many many unnecessary requests when calculating bug suggestions. * Give people an extra reason to break muscle memory and start using Treeherder, since summaries will take a couple of seconds longer to view than at present.

We can always re-enable this in a matter of minutes, should sheriffs need to go back to using TBPL for some reason.

Comment on attachment 8537885 [details] [diff] [review] Disable log prefetching Review of attachment 8537885 [details] [diff] [review]: ----------------------------------------------------------------- r++++

Ty :-) remote: https://hg.mozilla.org/webtools/tbpl/rev/99e0b27fd3a1