Message boards : ATLAS application : Network or Server Problems at CERN?
Message board moderation

To post messages, you must log in.

AuthorMessage
computezrmle

Send message
Joined: 15 Jun 08
Posts: 577
Credit: 5,984,047
RAC: 15,509
Message 32363 - Posted: 9 Sep 2017, 6:46:58 UTC
Last modified: 9 Sep 2017, 6:52:20 UTC

Last night and this morning my ATLAS VMs downloaded lots of data from the spare server ccfrontier.in2p3.fr.
This usually points out a problem with the main servers or the CERN network and may be checked.

<edit>Or a problem with lcgft-atlas.gridpp.rl.ac.uk</edit>
ID: 32363 · Report as offensive     Reply Quote
computezrmle

Send message
Joined: 15 Jun 08
Posts: 577
Credit: 5,984,047
RAC: 15,509
Message 32377 - Posted: 10 Sep 2017, 7:42:07 UTC

It still looks like lcgft-atlas.gridpp.rl.ac.uk can't deliver reasonable input for the WUs.
In addition the spare servers like ccfrontier.in2p3.fr and atlasfrontier-ai.cern.ch are very slow.
As a result the VMs need 25 minutes instead of the normal 5 minutes to finish the initial downloads.

@ David
Are you aware of that?
ID: 32377 · Report as offensive     Reply Quote
David Cameron
Project administrator
Project developer
Project scientist

Send message
Joined: 13 May 14
Posts: 176
Credit: 4,011,446
RAC: 9,285
Message 32384 - Posted: 11 Sep 2017, 9:52:44 UTC

There were some problems with the server at lcgft-atlas.gridpp.rl.ac.uk over the weekend, caused by overload from other non-ATLAS@Home tasks. I'm surprised that the failover servers are slow since they should basically be the same, but perhaps they were also overloaded due to the failing server.

Note this server is in the UK (and the ccfrontier.in2p3.fr server is in Lyon, France) so the problems are not related to CERN network or servers this time.
ID: 32384 · Report as offensive     Reply Quote
computezrmle

Send message
Joined: 15 Jun 08
Posts: 577
Credit: 5,984,047
RAC: 15,509
Message 32421 - Posted: 14 Sep 2017, 16:08:17 UTC

@ David Cameron

Just to make you aware.
According to DNS and my local squid log ATLAS normally uses 5 servers behind lcgft-atlas.gridpp.rl.ac.uk.
2 of them (lcgvo-frontier04.gridpp.rl.ac.uk and vm6.nubes.stfc.ac.uk) are currently down or again overloaded.
ID: 32421 · Report as offensive     Reply Quote
computezrmle

Send message
Joined: 15 Jun 08
Posts: 577
Credit: 5,984,047
RAC: 15,509
Message 33807 - Posted: 12 Jan 2018, 15:33:34 UTC

@ David Cameron

As I already wrote 1 of the 4 proxies (lcgvo-frontier04.gridpp.rl.ac.uk) that share the load for lcgft-atlas.gridpp.rl.ac.uk had (and still has) delivery problems during the last months.

As of this afternoon a 2nd one (lcgvo-frontier03.gridpp.rl.ac.uk) also started to report errors.
This should be investigated by your project partners in the UK.

Thanks in advance
ID: 33807 · Report as offensive     Reply Quote

Message boards : ATLAS application : Network or Server Problems at CERN?


©2018 CERN