Hi, I currently have four OSS''s that are connected to DDN storage. Each OSS can see all 24 DDN LUN''s but it use only six of them. I would like to remove one OSS and then mount 6 OST''s that belonged to OSS which had been removed on three remain OSS''s, 2 LUNS per OSS. My question is if that will actually work? I assume I would have to do only --writeconf on each storage device is that right? Regards Wojciech Turek -------------- next part -------------- An HTML attachment was scrubbed... URL: http://mail.clusterfs.com/pipermail/lustre-discuss/attachments/20070815/cfcebf9b/attachment.html
Wojciech, That is correct, what you would need to do is (assuming you are using lustre 1.4.x) modify your lmc script such that those luns (OST''s) are associated with the new OSS and generate the XML. Using that new XML you need to use it with lconf --writeconf. Are you using device- mapper-multipath ? if so you would need to make sure that the primary paths are from the controller that owns those LUN''s for optimal write performance. Please let us know if you have any other question. Thanks Anand On Aug 15, 2007, at 6:57 AM, Wojciech Turek wrote:> Hi, > > I currently have four OSS''s that are connected to DDN storage. Each > OSS can see all 24 DDN LUN''s but it use only six of them. I would > like to remove one OSS and then mount 6 OST''s that belonged to OSS > which had been removed on three remain OSS''s, 2 LUNS per OSS. My > question is if that will actually work? I assume I would have to do > only --writeconf on each storage device is that right? > > Regards > > Wojciech Turek > > > > > _______________________________________________ > Lustre-discuss mailing list > Lustre-discuss@clusterfs.com > https://mail.clusterfs.com/mailman/listinfo/lustre-discuss--- Anand Bisen Sr. Systems Architect, DataDirect Networks 9351 Deering Avenue, Chatsworth, CA 91311 Tel: 818-700-4053 Fax: 818-700-7665 Cell: 818-723-9752 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://mail.clusterfs.com/pipermail/lustre-discuss/attachments/20070815/bff547c3/attachment.html
Wojciech, That is correct, what you would need to do is (assuming you are using lustre 1.4.x) modify your lmc script such that those luns (OST''s) are associated with the new OSS and generate the XML. Using that new XML you need to use it with lconf --writeconf. Are you using device- mapper-multipath ? if so you would need to make sure that the primary paths are from the controller that owns those LUN''s for optimal write performance. Please let us know if you have any other question. Thanks Anand On Aug 15, 2007, at 6:57 AM, Wojciech Turek wrote:> Hi, > > I currently have four OSS''s that are connected to DDN storage. Each > OSS can see all 24 DDN LUN''s but it use only six of them. I would > like to remove one OSS and then mount 6 OST''s that belonged to OSS > which had been removed on three remain OSS''s, 2 LUNS per OSS. My > question is if that will actually work? I assume I would have to do > only --writeconf on each storage device is that right? > > Regards > > Wojciech Turek > > > > > _______________________________________________ > Lustre-discuss mailing list > Lustre-discuss@clusterfs.com > https://mail.clusterfs.com/mailman/listinfo/lustre-discuss--- Anand Bisen Sr. Systems Architect, DataDirect Networks 9351 Deering Avenue, Chatsworth, CA 91311 Tel: 818-700-4053 Fax: 818-700-7665 Cell: 818-723-9752 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://mail.clusterfs.com/pipermail/lustre-discuss/attachments/20070815/4eb2159d/attachment.html