Hi Ken,
----- Original Message -----
> I'm down to the onesies and twosies now in the QA failures (which is
> good), but this means I'm now looking at tests that have probably been
> failing on a host or a couple of hosts (out of 30+) for a long time.
>
> I've attached the .out.bad and .full files for two different failure
> signature for qa/823.
Lukas and I have been scratching our heads over the "unknown user" one
for awhile now - its being seen on one of the buildbots. The second
case I've never seen before though.
> Any suggestions on how to diagnose/fix these would be most appreciated.
I think these may be SASL regressions, but we'll need to be sure... I've
not seen them locally as yet. We may need notrun guards against certain
SASL versions (pmconfig -L has that info - but, ick). :(
In terms of debugging, my plan for the 'unknown user' case has been to
step through libsasl in gdb and see where it falls over with the input
provided to it in the test case... but, I've not got there yet.
cheers.
--
Nathan
|