I support going ahead with this proposal and will coordinate process-
wise with Morgen, Bear and QA.
On Jun 14, 2006, at 1:43 PM, Katie Capps Parlante wrote:
In the bug comments, Jared had two other suggestions:
- I suggest turning off multiple-retries in Chandler in favor of
longer
timeouts (per the 120 second change already). We're not helping
the server out
by asking it to start reports we're not going to wait for; this
behavior is a
recipe for server crunches because of the potential vicious loop
of never
finishing a report. Also, if we bump to 120 second timeouts and
that's done 3
times...ick. Let the user drive retries for now.
- Without Cosmo PROPFIND performance improvements in-hand, I
suggest the
Chandler timeout should be increased from even the 120 seconds, to
probably
300. We're at about 80 seconds now with no server load; under
even modest
server loads, PROPFINDs for a Mitch-3 sized calendar could exceed
120. I'm
going on the assumption that connection drops, connection refused,
etc are
dealt with as separate exceptions and not subject to the timeout.
If the
connection is open and it's a PROPFIND operation we're doing, we
should give
the server plenty of time to finish. If we get an order of magnitude
improvement, we can reduce this timeout significantly.
Thoughts about these proposals?
I created a bug for patching alpha2, to distinguish it from the bug
tracking the root problem.
https://bugzilla.osafoundation.org/show_bug.cgi?id=6050
Cheers,
Katie
Morgen Sagen wrote:
The patch is sitting in this comment:
https://bugzilla.osafoundation.org/show_bug.cgi?id=5996#c1
Also, that patch is already applied to the trunk.
On Jun 14, 2006, at 8:55 AM, Philippe Bossut wrote:
Ted Leung wrote:
On Jun 12, 2006, at 2:07 PM, Katie Capps Parlante wrote:
The proposal is that we go ahead and patch alpha2, spinning a
new build to give to dogfooders.
+1
+1 too. Does Morgen have a patch available? If so, that shouldn't
be hard for Bear to build the bits.
Just to make sure we follow due process though, someone should
create a bug, assign to Morgen, get the patch attached and
reviewed and the Bug Council approving it (just so that we all
see the extend of the patch and test required before it gets in).
Cheers,
- Philippe
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
Open Source Applications Foundation "chandler-dev" mailing list
http://lists.osafoundation.org/mailman/listinfo/chandler-dev
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
Open Source Applications Foundation "chandler-dev" mailing list
http://lists.osafoundation.org/mailman/listinfo/chandler-dev
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
Open Source Applications Foundation "chandler-dev" mailing list
http://lists.osafoundation.org/mailman/listinfo/chandler-dev
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
Open Source Applications Foundation "chandler-dev" mailing list
http://lists.osafoundation.org/mailman/listinfo/chandler-dev
