Hi, we have a problem with lustre-1.6.0b7 installation where two OSTs are running on one server but only one of the OSTs is in state FULL and the other in DISCONN. Both are successfully mounted and the recovery_status is "Complete". cat /proc/fs/lustre/osc/chicfs-OST0002-osc/ost_server_uuid chicfs-OST0002_UUID DISCONN cat /proc/fs/lustre/osc/chicfs-OST0000-osc/ost_server_uuid chicfs-OST0000_UUID FULL Is there a way to start a connection attempt again? Best Regards, Frank -- Dipl.-Inf. Frank Mietke | Fakult?tsrechen- und Informationszentrum Tel.: 0371 - 531 - 35538 | Fak. f?r Informatik Fax: 0371 - 531 8 35538 | TU-Chemnitz Key-ID: 60F59599 | frank.mietke at informatik.tu-chemnitz.de
On Sep 27, 2007 14:47 +0200, Frank Mietke wrote:> we have a problem with lustre-1.6.0b7 installation where two OSTs are running > on one server but only one of the OSTs is in state FULL and the other in > DISCONN. Both are successfully mounted and the recovery_status is "Complete". > > cat /proc/fs/lustre/osc/chicfs-OST0002-osc/ost_server_uuid > chicfs-OST0002_UUID DISCONN > > cat /proc/fs/lustre/osc/chicfs-OST0000-osc/ost_server_uuid > chicfs-OST0000_UUID FULLYou should be able to do "lctl --device %chicfs-OST0002-osc recover" Cheers, Andreas -- Andreas Dilger Principal Software Engineer Cluster File Systems, Inc.
Hi, On Thu, Sep 27, 2007 at 11:40:13AM -0600, Andreas Dilger wrote:> On Sep 27, 2007 14:47 +0200, Frank Mietke wrote: > > we have a problem with lustre-1.6.0b7 installation where two OSTs are running > > on one server but only one of the OSTs is in state FULL and the other in > > DISCONN. Both are successfully mounted and the recovery_status is "Complete". > > > > cat /proc/fs/lustre/osc/chicfs-OST0002-osc/ost_server_uuid > > chicfs-OST0002_UUID DISCONN > > > > cat /proc/fs/lustre/osc/chicfs-OST0000-osc/ost_server_uuid > > chicfs-OST0000_UUID FULL > > You should be able to do "lctl --device %chicfs-OST0002-osc recover"thanks for your answer. We found that the RAID controllers on the OSS had another RAID setup than before. This led to the behavior that some OSTs changed their physical OSS and thus wasn''t reachable for the MDS. Best Regards Frank> > Cheers, Andreas > -- > Andreas Dilger > Principal Software Engineer > Cluster File Systems, Inc. > >-- Dipl.-Inf. Frank Mietke | Fakult?tsrechen- und Informationszentrum Tel.: 0371 - 531 - 35538 | Fak. f?r Informatik Fax: 0371 - 531 8 35538 | TU-Chemnitz Key-ID: 60F59599 | frank.mietke at informatik.tu-chemnitz.de