Re: [Condor-users] Error from starter, jobs put on hold

[Dan Bradley <dan@xxxxxxxxxxxx>]

Michael Hess wrote:

> every day a couple (around 20) of jobs from our central submitter are put on hold.
> condor_q -l says:
>
> LastHoldReason = "Error from starter on pc-name.ourlocalnetwork.plymouth.ac.uk:
> STARTER failed to receive file(s) from <x.x.x.x:19086> Download acknowledgment
> missing attribute: Result"
> LastHoldReasonCode = 11
> LastHoldReasonSubCode = 0
>  

This error is rather unexpected.  Todd's suggestion of using 
PeriodicRelease will let you work around the problem, but I am very 
curious what is causing it.  Honestly, the only thing I can think of 
that would cause this to happen sporadically, as you describe, is some 
kind of memory or network data corruption.  However, it may also be that 
we are not correctly detecting some other "normal" network error, such 
as closing of the socket while the final transmission is in progress.  
Either way, I'd like to figure out what is going on.

When 6.8.4 comes out, there will be some extra information in the 
StarterLog accompanying this error message.  This should help us see 
whether the "download acknowledgment" contains corrupted data.

--Dan