[PRad] [Primex2] Fwd: [Clas_offline] Fwd: Re: /work/clas12 is over limit

Harut Avakian avakian at jlab.org
Thu Oct 26 17:53:53 EDT 2017


On 10/26/17 5:35 PM, Maxime Levillain wrote:
> I have cleaned up the space that I use on /work and I copied (copying) /work/hallb/prad/replay/raw and work/hallb/prad/replay/event_sel which are the latest production of our data (critical data) to the /cache/hallb/prad/production/new_format.
> I think Weizhi started also to clean up his space, but there is still 4To in there.
>
> Regards,
> Maxime Levillain
Hi Maxime,
If you copy files, then please remove them from the disk. In principle 
the cc script will copy to cache (means they appear on tape later on) 
all leftover files
in the first sweep it will move all files over 2 GB.  If we still stay 
above 120TB, it will sweep again at 1.2 GB and then 720 MB (each sweep 
all or nothing).
That means if you copy large files to somewhere on cache and leave them 
on the disk, they will be copied to cache again.

Best,
Harut
>
> ----- Original Message -----
> From: "Eugene Pasyuk" <pasyuk at jlab.org>
> To: "prad" <prad at jlab.org>, "primex2" <primex2 at jlab.org>
> Sent: Thursday, October 26, 2017 3:40:37 PM
> Subject: [Primex2] Fwd: [Clas_offline] Fwd: Re: /work/clas12 is over limit
>
> See message below.
>
> There are some changes coming with /work disk related to its migration to different server. I affects not only /work/clas/... but also /work/hallb/... It needs to be cleaned up. Whatever is not needed must be deleted. Whatever we need to keep needs to be moved somewhere else (/volatile for instance). Something that needs to be kept for a long term can be copied to tape.
>
> Send any comments and questions you may have to Harut.
>
> -Eugene
>
>
> From: "Harut Avagyan" <avakian at jlab.org>
> To: "clas offline" <clas_offline at jlab.org>
> Sent: Thursday, October 26, 2017 10:25:17 AM
> Subject: [Clas_offline] Fwd: Re: /work/clas12 is over limit
>
>
>
> Dear All,
>
> As announced before we are moving to new work disks and all data that is still there on the hallb work disks above the quota will be moved to tapes. We will have a meeting with Chip today to
>
>
> discuss technical details of how that will happen.
>
> Please send me any comments you have on that.
>
> Regards,
>
> Harut
>
>
> -------- Forwarded Message -------- Subject: 	Re: /work/clas12 is over limit
> Date: 	Thu, 26 Oct 2017 10:08:06 -0400
> From: 	Chip Watson <watson at jlab.org>
> To: 	Sandy Philpott <philpott at jlab.org> , Harut Avakian <avakian at jlab.org> , Vardan Gyurjyan <gurjyan at jlab.org> , Veronique Ziegler <ziegler at jlab.org> , clas12 at jlab.org
>
> All,
>
> We will start today to perform automatic shrinking of /work/hallb
> (currently at 125 TB).  We have two different approaches we can use to
> accomplish this, and the choice of which of these two will be active by
> COB today is up to you (Hall B).  I am free to discuss these two choices
> with you at either 11:30 today or 1:30 today.  As many as are interested
> in this quick conversation can drop by my office at either of those two
> times (hm... limit 6, please).  Implementation starts at 2 p.m. and will
> be running by 3:30 p.m. (we don't like making changes on Friday).  Both
> are simple, and both involve moving files to tape.  It is only a
> question of which files (big or old).
>
> Once the /work file server is upgraded (by mid-January at the latest),
> its quota system will be the manager of all of /work, and the process
> turned on today will no longer be used at that time. This is just a
> temporary fix to ensure that one hall's activities do not impact the
> rest of the lab.
>
> thanks,
>
> Chip
>
>
> On 10/26/17 9:15 AM, Sandy Philpott wrote:
>> Hi Harut, Vardan, Veronique, CLAS12,
>>
>> Please note that the /work/clas12 tree is now at Lustre quota, with 48TB used; this will severely limit the ability of CLAS12 to use any ENP batch farm resources.  See usage at https://scicomp.jlab.org/scicomp/index.html#/work >
>> For large files, please delete them, or move them to /cache (for going to tape) or /volatile  (scratch).
>>
>> Regards,
>> Sandy
>>
> _______________________________________________
> Clas_offline mailing list
> Clas_offline at jlab.org
> https://mailman.jlab.org/mailman/listinfo/clas_offline
>
> _______________________________________________
> Primex2 mailing list
> Primex2 at jlab.org
> https://mailman.jlab.org/mailman/listinfo/primex2


-- 
___________________________________________________________________________
Harut  Avagyan                           e-mail  - avakian at jlab.org
12000 Jefferson Ave. Suite 5             tel/fax - 1(757)269-7764/5800
Newport News, VA 23606                   http://www.jlab.org/~avakian



More information about the PRad mailing list