These are chat archives for cern-eos/eos/operations

2nd
Jun 2016
I've had this appear, if I go direct to the FSTs, and find the files on disk, their shasums match
but I can't read the file via either xroot protocol or via eosd
there was no write to this file, only thing that happened was opening it in vi, and then exiting
David Jericho
@davidjericho
Jun 02 2016 08:47
if I got to the FST, find the file, copy it off fs, delete the filevia the eos cli tool, then add the file again,e verything is fine
David Jericho
@davidjericho
Jun 02 2016 13:21
@apeters1971 also, I have noticed that at least in >0.3.182, node ls -m, show's empty values for the eos...version and other system fields
@apeters1971 and I'm getting CRIT errors in the logs for missing fmd for /eos/instancename/proc/master
none of these seem to affect the system
Andreas-Joachim Peters
@apeters1971
Jun 02 2016 14:39
@davidjericho Hi David, I am a little bit confused by the crossing issues, but do you have eventually problems with a DNS entry? Why did the master name change? old-manager="actn-s1.cdn.aarnet.edu.au" new-manager="dksn-s1.cdn.aarnet.edu.au"
Considering the node ls -m, I don't have that, I have the impression, that you might have different instance names between master and slave or some other configuration problem ...
Andreas-Joachim Peters
@apeters1971
Jun 02 2016 14:44
@davidjericho Because, a missing FMD for /eos/instancename/proc/master in the FMD log is absolutely weird, this means that a slave was writing to an old proc file as a file (instance name change or similiar)
David Jericho
@davidjericho
Jun 02 2016 23:07
@apeters1971 I decided to restart both mgms, and it seems to have come good
@apeters1971 the master host did change, but the config and hostnames never changed,