Re: CVS, replication, and triggers (oh my!)

[Mark J]

>From what I've read, this issue is not fully understood -- it's hard
to replicate the broken pipe error.  Our loginfo is pretty vanilla as
well:
DEFAULT (echo ""; echo "Date: " 'date'; cat) |
$CVSROOT/CVSROOT/scripts/bugzilla-watcher default $USER %(sVv)

I don't know where the failue is; at the trigger, in bugzilla-watcher,
or after bugzilla-watcher returns exit 0 back to the trigger.

I'll play around a little more and see if I can't get it to break
again with the same (similar) transaction.

Thanks,
Mark 

On Sat, 23 Dec 2006 14:24:11 -0500 (EST), address@hidden
(Larry Jones) wrote:

>Mark J writes:
>> 
>> We have a replicated repository between MN and India.  We are using
>> WANdisco.  And, cocasionally, we are witness to the mysterious power
>> of the Broken Pipe Error (BPE) as caused by loginfo not reading all
>> the log text.
>[...]
>> I'd appreciate any thoughts.
>
>Fix your loginfo script so that it *does* read all the log text.  If
>it's not the right hostname, do ``cat >/dev/null'' before you exit.
>
>-Larry Jones
>
>That's the problem with nature.  Something's always stinging you
>or oozing mucus on you. -- Calvin
>