Running Fsck Online - IBM TS7650G User Manual

Running fsck online

ProtecTIER Version 3.3.7 includes the capability of running fsck online. You do not
need to take the system offline while you run fsck. The system stays operational so
that backups and restores continue while you check the consistency of the file
systems.
Before you begin
Before you run fsck, run the procedure in "Checking and repairing errors" on
page 173 to check for errors that require the use of fsck. Also, ensure that the file
system on which you intend to run fsck is unmounted.
To run the Online fsck Recovery, your system must meet the following
requirements:
v ProtecTIER must be configured as a 2-node cluster.
v Each node in the ProctecTIER cluster must be configured with at least 64 GB of
RAM.
v One of the following criteria must be true:
– Medium changer is active on the node that is up during the FSCK Recovery
process, OR
– Control Path Failover (CPF) is enabled if you use Tivoli Storage Manager as
its backup application.
If your system does not meet this criteria, use the procedure that is described in
"Running fsck" on page 176.
Procedure
1. Select one of the ProtecTIER nodes.
2. At the login: prompt, log in with the ID root and the password admin.
3. Shut down the vtfd and ptcluster services in the following order:
# service vtfd shutdown
# service ptcluster shutdown
Attention: Make sure to use the shutdown option and not the stop option. If
you use stop, the services restart automatically if a reboot occurs.
4. Start the cman and clvmd services in the following order:
# service cman start
# service clvmd start
5. For each of the problematic file systems, run the gfs_fsck –yv command in the
background with high verbosity and direct the output to a file:
# gfs_fsck -yv logical volume name > /pt_work/vgxx_fsck.out 2>&1 &
For example, the following command runs gfs_fsck on the logical volume
/dev/mapper/vg0-lv_vg0:
# gfs_fsck -yv /dev/mapper/vg0-lv_vg0 > /pt_work/vg0_fsck.out 2>&1 &
The output is saved in the file "/pt_work/vg0_fsck.out."
Note: ProtecTIER supports up to 10 parallel gfs_fsck instances.
6. When all the gfs_fsck processes are complete, remove the /etc/pt/
dont_start_vtfd and /etc/pt/dont_start_ptcluster.
# rm –f /etc/pt/dont_start_vtfd
# rm –f /etc/pt/dont_start_ptcluster
Chapter 17. Troubleshooting
175