launchpad-dev team mailing list archive

[Robert Collins <robertc@xxxxxxxxxxxxxxxxx>]

Today was a pretty routine day, performance wise.

I got 3 timeout bugs fixed, found a regression in another page that
isn't (yet) causing timeouts and filed a bug on it. William found a
xss hole, fixed it and got that deployed.

We had a moderate jump in timeouts in the most recent oops report:
* 415 Exceptions
 * 1112 Time Outs
 * 1622 Soft Time Outs

he top few are: (F means fixed)
    Hard / Soft  Page ID
     183 /  130  BugTask:+index
F      81 /  123  Distribution:+bugtarget-portlet-bugfilters-stats
      79 /   91  Distribution:EntryResource:searchTasks
      70 /  290  Archive:+index
F      48 /   69  ScopedCollection:CollectionResource:#person-page-resource
F      46 /   93  BranchSet:CollectionResource:#branches
F      41 /   29  ProductSet:CollectionResource:#projects
      36 /   14  DistroSeries:+queue
      33 /    9  POFile:+translate
      33 /    6  Cve:+index

The fixed ones alone will drop us back under 1000 a day, once we
deploy (which we can only do once the xss hole has passed buildbot and
can be the next revision deployed).

No particular lessons today, other than to note that
lp.services.database.bulk.load_related seems to work pretty well; if
we can get a more general version that supports dealing with
tuple-inputs and chaining subsequent bulk loads, we might be able to
get rid of many of the pre_iter_hook local function definitions (but
not all).

-Rob