I spoke to raja at the tier 1 liason meeting today re T2Ds for LHCb and ipV6ing.
His response is that T2Ds (not T2s) are in the same cateogory as T1s in that LHCb would like more space dual stacked ASAP, They do nto have a requirement that all storage at T1/T2Ds to be dual stacked.
Brian
-----Original Message-----
From: GRIDPP2: Deployment and support of SRM and local storage management [mailto:[log in to unmask]] On Behalf Of Brian Davies
Sent: 15 March 2017 09:47
To: [log in to unmask]
Subject: Re: Agenda for tomorrow's storage meeting
(Monday 1500)
Here they are for the group.
Minutes Storage Meeting 08/02/2017
ZFS--
In production at ECDF Markus E upgraded in a rolling manner old servers into using ZFS.
New disk servers at Glasgow will have it.
(allows for resizing of file systems on the fly Durham use it locally.
Lancaster testing ( Matt was not there so do not know if any is in production yet.)
IPV6--
Didn’t really discus. May want to discuss my email to the list and the following discussion of it. Taking in news from WLCG GDB may be useful.
Accounting--
No real comments about use of bdii or json as preferable.
Related is QMUL issue with SRM space token usage not being updated when files are deleted (and probably written) via WebDAV.
Believed only to be an issue for sTorM+Lustre and not SToRM+GPFS. )( GPFS has a backend approach which auto fixes.) possible Lustre fix might exist but QMUL does not have that functionality.
Scaling issue for QMUL for use of WebDAV. QMUL can have multiple gridftp severs ( and currently does,) but only has the capability to have one WebDAV server. Multi WebDAV server functionality is planned but no yet available.
Site Reports--
Glasgow:
Playing new storage jenga.( needs to drain old servers before they can be physically removed so as to make room for new servers to be installed, but only ( current ) available space is on the new servers which can't be plugged in...
setting up XROOTD federation test point. Sam getting back to work on arch-caching for the T2c taskforce work.
Edinburgh: busy with "training form Markus" . Installing fresh test DPM on clean machine.
Raul: is testing XROOTD cache with Sam S. Plan is to deploy new storage into this instance.
Winnie: "HFDS works well when it does"
Markus: is testing a plain XROOTD redundant service. ( test setup.)
Jon B: (Liverpool) has new storage to rack up. (storage "Jenga" s also needed here.) Plan these servers to be DPM1.9/Centos7/ZFS. Then will look to upgrade headnotes to similar.
Govind: How much space in the new space token do snoplus need? (IF as we believe) the space is only really used for job output it doesn't need to be that muck?? need to discuss with snoplus.
Dan T: (QMUL) . Discussing with vendors for new storage to go into Lustre.
-----Original Message-----
From: GRIDPP2: Deployment and support of SRM and local storage management [mailto:[log in to unmask]] On Behalf Of Jensen, Jens (STFC,RAL,SC)
Sent: 14 March 2017 18:49
To: [log in to unmask]
Subject: Agenda for tomorrow's storage meeting
0. Operational blog posts
Hurrah for JohnB and Brian who have posted blog, er, posts. Why are there only two on the front page?
Also, any notes from last week? Brian said he'd send me some...
1. If David joins us, perhaps we can have a storageanddatamanagementological summary of IGSC? If not, we will have to wait till Rob A gets back, but he's extending his holiday with a trip to Japan while he's in the neighbourhood.
There was also a GDB: http://indico.cern.ch/event/578984/ although it seems to be mostly computing focused.
2. Hmm, we still have some keydocs that could do with a bit of an update. To name a few, those of StoRM and DPM (from 2014), and dCAche and suggestions for hardware are reporting a measly 15-20% completion...
While we are here, I note Matt's ticket handling guidelines which will be judged accurate in about three weeks' time ;-P
3. Ends, loose: accounting, ZFS; there was an SKA kick off in den Haag (the Hague); "small" VOs (LZ?), IPv6 storage(?), data transfers
4. If there's time perraps a word on the EGI/EUDAT/Indigo data infrastructure update?! and how they might be relevant to GridPP.
5. AOB
|