GFS 6.0 Trouble Shooting

1 Startreihenfolge der Dienste

1) rawdevices
2) pool
3) ccsd
4) lock_gulmd
5) gfs
6) clumanager

2 RAWDEVICES

raw -qa
service rawdevices status

3 POOL

pool_tool -s
service pool status
lsmod | grep pool

4 CCSD

ccs_read list
ccs_read file file.ccs
service ccsd status

5 LOCK_GULMD

gulm_tool servicelist localhost:core [hostname]
gulm_tool nodeinfo localhost:core hostname
gulm_tool nodelist localhost:core
gulm_tool getstats localhost:core
gulm_tool setverb localhost:core "ReallyAll, -Heartbeat"
gulm_tool setverb localhost:core "Default"
service lock_gulmd status
cat /proc/gulm/lockspace
lock_gulmd -v ReallyAll

6 GFS

service gfs status
mount -tgfs
df -tgfs
gfs_tool gettune <mountpoint>
gfs_tool settune <mountpoint> <parameter> <value>
gfs_tool jindex <mountpoint>

7 CLUMANAGER

grep config_viewnumber /etc/cluster.xml
clustat -x | grep config_viewnumber
clustat
service clumanager status

7.1 gulm_tool

dient zum konfigurieren und status auslesen des gulm_lockd
Status auslesen

gulm_tool getstats localhost:core

lockd-master auf slave verlagern (master ist immer der, der am lägsten online ist)

gulm_tool switchPending localhost:core # auf aktuellem master ausführen

8 LINKS

http://www.redhat.com/archives/linux-cluster/2006-January/msg00034.html http://sources.redhat.com/cluster/faq.html