[gpfsug-discuss] Placement policies and copies

[Simon Thompson (Research Computing - IT Services)]

Hi Marc,

Thanks. With the migrate option, does it remove the second copy if already present? Or do you still need to do an mmrestripefs to reclaim the space?

Related: if the storage pool has multiple failure groups, will GPFS place the data into a single pool, or will it spray the data over all NSD disks in all failure groups?

I think I'll stick to using a pool with NSD disks in a single failure group, so I know where the files are, but would be useful to know. I assume that if the pool then goes offline, I won't lose my whole FS, just not have access to the non replicated fileset?

Thanks

Simon
________________________________
From: gpfsug-discuss-bounces at spectrumscale.org [gpfsug-discuss-bounces at spectrumscale.org] on behalf of Marc A Kaplan [makaplan at us.ibm.com]
Sent: 30 November 2015 17:58
To: gpfsug main discussion list
Subject: Re: [gpfsug-discuss] Placement policies and copies

>From the Advanced Admin book:

File placement rules:

RULE [’RuleName’]
SET POOL ’PoolName’
[LIMIT (OccupancyPercentage)]
[REPLICATE (DataReplication)]
[FOR FILESET (’FilesetName’[,’FilesetName’]...)]
[WHERE SqlExpression]


So, use REPLICATE(1)

That's for new files as they are being created.

You can use mmapplypolicy and the MIGRATE rule to change the replication factor of files that already exist.

--marc of GPFS.



From:        "Simon Thompson (Research Computing - IT Services)" <S.J.Thompson at bham.ac.uk>
To:        gpfsug main discussion list <gpfsug-discuss at spectrumscale.org>
Date:        11/30/2015 11:27 AM
Subject:        [gpfsug-discuss] Placement policies and copies
Sent by:        gpfsug-discuss-bounces at spectrumscale.org
________________________________



Hi,

I have a file system which has the default number of data copies set to 2.

I now have some data Id like to have which only has 1 copy made. I know that files and directories don't inherit 1 copy based on their parent.

Can I do this with a placement rule to change the number of copies to 1?

I don't really want to have to find the file afterwards and fix up as that requires an mmrestripefs to clear the second copy.

Or if I have a pool which only has nsd disks in a single failure group and use a placement policy for that, would that work? Or will gpfs forever warn me that due to fs changes I have data at risk?

Thanks

Simon
_______________________________________________
gpfsug-discuss mailing list
gpfsug-discuss at spectrumscale.org
http://gpfsug.org/mailman/listinfo/gpfsug-discuss



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://gpfsug.org/pipermail/gpfsug-discuss_gpfsug.org/attachments/20151130/aa18dc33/attachment-0002.htm>