Mailing List Archives
Authenticated access
|
|
|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [HTCondor-users] Interpreting XferStatsLog and understanding message flow
- Date: Thu, 15 Mar 2018 14:42:37 -0500
- From: Todd Tannenbaum <tannenba@xxxxxxxxxxx>
- Subject: Re: [HTCondor-users] Interpreting XferStatsLog and understanding message flow
On 3/15/2018 2:09 PM, George Papadimitriou wrote:
Hello,
Are there any race conditions while condor is trying to write to
XferStatsLog ?
This could explain the weird log entries I observed.
Thanks,
George
Hi George,
There certainly is something strange going on. Certainly the job id
should not be different in the same record!
Thanks for reporting it, we are looking into it. Sure looks like you
found a bug. The good news is I checked a couple of production submit
hosts here at Univ of Wisconsin, and also observed similar weird
entries... so at least reproducing the problem does not involve
something specific in your environment, which should make it easier to fix.
I made a developer ticket about it here:
https://htcondor-wiki.cs.wisc.edu/index.cgi/tktview?tn=6608
Meanwhile, what does condor_version say? Also does appending the
following into your condor_config[.local] file on your submit machine
make any difference:
SHADOW_STATS_LOCK = $(LOCK)/XferStatsLock
LOCK_DEBUG_LOG_TO_APPEND = True
? (don't forget to do a condor_reconfig afterwards...)
If they don't make a difference, please recall to remove them (esp the
LOCK_DEBUG_LOG_TO_APPEND line... it should not need to be True, and
locking the log file before every write is a scalability liability).
regards,
Todd