[CalendarServer-users] Connection aborted - took too long to close

[Andre LaBranche]

> On Feb 3, 2015, at 8:21 PM, Jacques Distler <distler at golem.ph.utexas.edu> wrote:
> 
> 
>> On Feb 3, 2015, at 6:07 PM, Andre LaBranche <dre at apple.com> wrote:
>> 
>> Hard to say what's up without seeing more information (such as any interesting error.log content),
> 
> Well, that's the thing.
> 
> There ARE NO other errors except for the
> 
>  [twext.web2.channel.http#error] Connection aborted - took too long to close: ...
> 
> errors. So I don't know what exactly is timing out.

Try setting DefaultLogLevel to ‘info’ in caldavd.plist, then bounce the service or HUP the master process.

-dre

> 
>> however there's a reasonably small update to 5.4-dev that has some key fixes made since 5.3.
>> 
>> http://trac.calendarserver.org/log/CalendarServer/branches/release/CalendarServer-5.4-dev
>> 
>> There is one known issue in 5.3 regarding how we handle updating calendar event data when the format changes (as it has recently). In 5.3, the approach is to leave the data in the old format until it is requested, and then update it on the fly as it is vended to the client. This fails when there are enough events being 'touched' by the request that the processing time for doing the format upgrades exceeds the client's request timeout threshold. This is fixed in 5.4-dev.
> 
> Well, that sounds like a reasonable culprit.
> 
> So I "upgraded" from 5.3 to 5.4-Dev (revision 14365). The
> 
>  [twext.web2.channel.http#error] Connection aborted - took too long to close:...
> 
> errors seem to be unaffected. Again, the server seems to start up normally, and no other errors appear in the error.log file.
> 
> 
>> Be sure to have backups before you do anything :)
>