GridPP Technical Meeting - Diskless Sites

Friday 27 Mar 2020, 11:00 → 12:50 Europe/London

Virtual Only

Alastair Dewhurst (Science and Technology Facilities Council STFC (GB)), Andrew McNab (University of Manchester), David Colling (Imperial College Sci., Tech. & Med. (GB))

Weekly meeting slot for technical topics. We will try and focus on one topic per meeting.  We will announce at the Tuesday Ops meeting if this meeting is going ahead and if so the topic to be discussed.

11:00 → 11:45 Discussion on Diskless Sites

Speakers: Alastair Dewhurst (Science and Technology Facilities Council STFC (GB)), Dr Teng LI (University of Edinburgh, UK)

DisklessSites20200327.pdf

Teng went through his slides that he had presented at the DOMA Access meeting earlier in the week:

https://indico.cern.ch/event/895815/

 

- We never resolve what exactly caused network saturation between Birmingham and Manchester

This has been used as a major justification for the need for an XCache.  It may not have been caused by HEP traffic.

 

- The monitoring shows that certain files are extremely popular and we should understand why.

One theory is that the popular files are all the HC test jobs.  Analysis input tar files were also common.

 

Actions:

 

ECDF and RAL to compare their settings for XRootD monitoring into Elastic Search and to look at making the plots public.

 

Duncan to circulate link to JISC network plots (Alastair will try signing up)   https://www.jisc.ac.uk/forms/apply-to-use-netsight3

 

ATLAS have the ability to configure I/O for a Panda Queue.  ATLAS are making changes to increase I/O for some sites.  Teng will monitor this over the next few weeks and report to the OPs meeting if problems appear.  RAL to watch Sheffield traffic.

 

Sam Skipsey will see if anyone is doing testing intelligent caching.

 

11:45 → 12:00 AoB