There appears to be a dodgy bdii node at RAL.
I get the following hosts behind the DNS alias:
walker@heppc300:~$ nslookup lcg-bdii.gridpp.ac.uk
Server: 138.37.6.1
Address: 138.37.6.1#53
Non-authoritative answer:
lcg-bdii.gridpp.ac.uk canonical name = lcgbdii.gridpp.rl.ac.uk.
Name: lcgbdii.gridpp.rl.ac.uk
Address: 130.246.183.72
Name: lcgbdii.gridpp.rl.ac.uk
Address: 130.246.183.73
Name: lcgbdii.gridpp.rl.ac.uk
Address: 130.246.183.69
Name: lcgbdii.gridpp.rl.ac.uk
Address: 130.246.183.70
And if I explicitly choose one of them, it gives no results for the query:
walker@heppc300:~$ export
LCG_GFAL_INFOSYS=130.246.183.70walker@heppc300:~$ for ((i=1;i<=10;i++));
do lcg-infosites --vo snoplus.snolab.ca wms |wc -l ; done
0
0
0
0
0
0
0
0
0
0
But the others are fine. For example this one:
walker@heppc300:~$ export LCG_GFAL_INFOSYS=130.246.183.69
walker@heppc300:~$ for ((i=1;i<=10;i++)); do lcg-infosites --vo
snoplus.snolab.ca wms |wc -l ; done
6
6
6
6
6
6
6
6
6
6
Last time I had this problem, it turned out to be a problem with the
upstream information (one of a loadbalanced pair of information
providers was sending the wrong information) rather than the bdii. I
think this was discussed in
https://ggus.eu/ws/ticket_info.php?ticket=82794 - but as GGUS is down at
the moment I can't check (which is partly why I'm sending this to
tb-support).
QMUL is seeing regular job failure for ATLAS - and this is a likely cause.
Chris
|