Hello chaps and chappettes,
I've got a bit of a mystery here today involving Lancaster repeatedly
failing the ops replica manager sfts. Checking the error messages they
were having permission denied problems. I followed it up, and sure
enough the directory /pnfs/lancs.ac.uk/data/ops/generated/2006-12-07/
was owned by root:root. I manually switched it to the correct
permissions, which seems to make the sft jobs happier, and now I'm
trying to figure out why such an oddity happened.
Checking the billing db and logs it seems like there was no access by
ops prior to the permission change this afternoon. Checking my other
logs lead me to mystery number 2- none of my log files are being
filled since I restarted dcache on Tuesday, which is quite worrying.
As far as I can see there's nothing that I changed that would have
effected things (I was going to change which directory the billing
logs were sent to, but in the chaos of the AC breakdown I forgot to
set it).
So I have 2 mysteries here, and no easy way to check what happened to
cause them. Anyone seen similar phenomena? Brian reckons there have
been sightings of pnfs directories being created with the root
permissions before (perhaps at RAL?) but I'm afraid I couldn't track
the case down. The empty log files are particular worrying, in fact
they're the stuff of sysadmin nightmares. But other then restarting
dcache and hoping (which for one would annoy the 20 or so people
transfering files at the moment) I'm not sure what to do!
cheers,
Matt, feeling quite dramatic today.
|