Re: [Lustre-discuss] Lustre locking up on login/interactive nodes
On Mon, 2008-07-21 at 12:04 -0400, Brock Palen wrote: Ok will keep in mind. Looks the same though, Indeed, in most cases it is the same, with the one exception that from syslog, we get time context. Its odd though, if I login to the same machine I can move to that directory list the files etc. read files on those OST's and notice this was eviction by the MDS, I see no lost network connections or network errors. Strange not good not good at all. The syslog data is the same, its below: Right. There is nothing terribly useful in it though. It's just the notification of an eviction. The real question is why did it get evicted. The evictor will know more about that than the evictee. If the evictor doesn't log anything more than a didn't hear from client ... in ... seconds, evicting then something is preventing the client from sending traffic to the evictor. b. signature.asc Description: This is a digitally signed message part ___ Lustre-discuss mailing list Lustre-discuss@lists.lustre.org http://lists.lustre.org/mailman/listinfo/lustre-discuss
[Lustre-discuss] Start with GigE; Move to IB
All: As we plan our Lustre rollout, we're contemplating using GigE for our initial topology and replacing that with IB six months to a year down the line. Can that be done without rebuilding the filesystem and associated metadata stores? Will that be a relatively straightforward replacement process or are we looking at a complicated changeover when we move to IB? Thanks, David -- David Bogen :: (608) 263-0168 Unix SysAdmin :: IceCube Project [EMAIL PROTECTED] ___ Lustre-discuss mailing list Lustre-discuss@lists.lustre.org http://lists.lustre.org/mailman/listinfo/lustre-discuss
Re: [Lustre-discuss] Start with GigE; Move to IB
On Tue, 2008-07-22 at 15:39 -0500, David Bogen wrote: As we plan our Lustre rollout, we're contemplating using GigE for our initial topology and replacing that with IB six months to a year down the line. Can that be done without rebuilding the filesystem and associated metadata stores? Will that be a relatively straightforward replacement process or are we looking at a complicated changeover when we move to IB? That can be done quite easily. You want to look for the changing a server NID entry in the manual. b. signature.asc Description: This is a digitally signed message part ___ Lustre-discuss mailing list Lustre-discuss@lists.lustre.org http://lists.lustre.org/mailman/listinfo/lustre-discuss
Re: [Lustre-discuss] Start with GigE; Move to IB
It's a pretty straightforward process. You may have to take the filesystem down for 15 minutes while you modify the nids but I think thats about it. I've done it with a 40tb filesystem and had no trouble. On Jul 22, 2008, at 4:39 PM, David Bogen wrote: All: As we plan our Lustre rollout, we're contemplating using GigE for our initial topology and replacing that with IB six months to a year down the line. Can that be done without rebuilding the filesystem and associated metadata stores? Will that be a relatively straightforward replacement process or are we looking at a complicated changeover when we move to IB? Thanks, David -- David Bogen :: (608) 263-0168 Unix SysAdmin :: IceCube Project [EMAIL PROTECTED] ___ Lustre-discuss mailing list Lustre-discuss@lists.lustre.org http://lists.lustre.org/mailman/listinfo/lustre-discuss ___ Lustre-discuss mailing list Lustre-discuss@lists.lustre.org http://lists.lustre.org/mailman/listinfo/lustre-discuss