I checked the last 48 hours of Tinderbox logs to find out why we are
seeing so much orange.

Green cycles: 635
Orange cycles: 44 (6%)
TOTAL:        679

Full cycle failures:   0
Quick cycle failures: 29
Perf cycle failures:  13
TOTAL:                42 (seems I missed two somewhere)

Quick cycle failures:

TestRecurrenceImporting (mac and linux): 16
Segfault (mac):                           7
TestCertstoreView:                        2
Timeout:                                  2
TestSharing (linux):                      1
recTestNewTask (linux):                   1
TOTAL:                                   29

Perf cycle failures:

Timeout:                                  7
TestRecurrenceImporting:                  5
TOTAL:                                   12

Quick + Perf TOTAL:                      41 (seems I missed one more)


TestCertstoreView failures were a known issue with the test (there's
probably a bug on that).

TestSharing failed with 500 server error, didn't dig deeper.

Various perf tests timed out; don't know why.

There is already a bug on the mac segfaults:
https://bugzilla.osafoundation.org/show_bug.cgi?id=8909

There is also a bug for TestRecurrenceImporting:
https://bugzilla.osafoundation.org/show_bug.cgi?id=8985


In my opinion we have too much failures right now. It makes reading
Tinderbox results very annoying, and the failure emails are almost
completely useless.

If 8909 and 8985 were fixed, we'd be down to about 8 failures per day,
or about 2% failure rate overall. I think would make the emails useful
again and any orange on a quick build almost certainly a bug caused by
the previous checkin, rather than a random failure.

-- 
  Heikki Toivonen


Attachment: signature.asc
Description: OpenPGP digital signature

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

Open Source Applications Foundation "chandler-dev" mailing list
http://lists.osafoundation.org/mailman/listinfo/chandler-dev

Reply via email to