[Lustre-discuss] Fwd: lctl dl -t on mds differes from reality :)

Brian O'Connor briano at sgi.com
Mon Jan 16 16:08:48 PST 2012

I sent this to wc-discuss, but forgot to cc lustre-discuss, sorry for 
the duplicates

    I have a oss failover pair that has 10 osts each, 20 osts total.

We had an IB melt down. After the IB glitches were fixed an lctl dl -t
on the mds tells me that all osts are on one server. i.e a failover

But the osses disagree. There is no corosync/pacemaker failover evewnt
in the logs and df and lctl dl on the oss show that the osts are
properly distributed over the osses ie the osses don't agree with the mds.

Is this likely to be fixed as clients connect and files are read , or is 
  there some thing else screwed.

Brian O'Connor
SGI Consulting
Email: briano at sgi.com, Mobile +61 417 746 452
Phone: +61 3 9963 1900, Fax:  +61 3 9963 1902
357 Camberwell Road, Camberwell, Victoria, 3124

More information about the lustre-discuss mailing list