[Linux-cluster] Clustering without DRBD?

[Junliang Li]

在 2013-11-20三的 17:55 -0700，Carlos Konstanski写道：
> I cannot find any documentation on how to do clustering WITHOUT DRBD. I
> have a huge expensive SAN, and I do not need replicated storage. It's
> already RAIDed across a kajillion spindles.
> 
> I'm going for a postgresql active/passive setup. I had my test working
> with DRBD. But when I try (without documentation) to remove DRBD from
> the setup, nothing fires up. I feel like I'm just missing something that
> triggers the filesystem/postgres into mounting/starting.

> 
> With DRBD (works):
> 
> crm
> cib new vip
> configure property stonith-enabled=false
> configure property no-quorum-policy=ignore
> configure rsc_defaults resource-stickiness=100
> configure primitive DBIP ocf:heartbeat:IPaddr2 params ip=192.168.122.100
> cidr_netmask=32 op monitor interval=30s
> configure show
> cib commit vip
> quit
> 
> crm
> cib new drbd
> configure primitive drbd0 ocf:linbit:drbd params drbd_resource=postgres
> op monitor interval=29s role=Master op monitor interval=31s role=Slave
> configure ms ms_drbd0 drbd0 meta master-max=1 master-node-max=1
> clone-max=2 clone-node-max=1 notify=true
> configure show
> cib commit drbd
> quit
> 
> crm
> cib new pgfs
> configure primitive fs_pgdata ocf:heartbeat:Filesystem params
> device=/dev/drbd0 directory=/var/lib/postgresql fstype=ext4
> configure primitive postgresql lsb:postgresql op monitor interval=30
> timeout=60 op start interval=0 timeout=60 op stop interval=0 timeout=60
> configure group group_postgres fs_pgdata DBIP postgresql
> configure colocation fs_on_drbd0 inf: group_postgres ms_drbd0:Master
> configure order pgdata_after_drbd0 inf: ms_drbd0:promote
> group_postgres:start
> configure location master-prefer-pgcluster1 DBIP 50: pgcluster1
> configure show
> cib commit pgfs
> quit
> 
> 
> 
> Without DRBD (doesn't work):
> 
> crm
> cib new vip
> configure property stonith-enabled=false
> configure property no-quorum-policy=ignore
> configure rsc_defaults resource-stickiness=100
> configure primitive DBIP ocf:heartbeat:IPaddr2 params ip=192.168.122.100
> cidr_netmask=32 op monitor interval=30s
> configure show
> cib commit vip
> quit
> 
> crm
> cib new pgfs
> configure primitive fs_pgdata ocf:heartbeat:Filesystem params
> device=/dev/sda directory=/var/lib/postgresql fstype=ext4
> configure primitive postgresql lsb:postgresql op monitor interval=30
> timeout=60 op start interval=0 timeout=60 op stop interval=0 timeout=60
> configure group group_postgres fs_pgdata DBIP postgresql
> configure order pgdata_after_san inf: fs_pgdata group_postgres:start
> configure location master-prefer-pgcluster1 DBIP 50: pgcluster1
> configure show
> cib commit pgfs
> quit
> 

Your configuration
"configure order pgdata_after_san inf: fs_pgdata group_postgres:start"
looks strange.
>From above, fs_pgdata is a member of group_postgres， maybe a deadlock ?
Try to delete this order setting, it already does what you want in group
rule.

Regards,
Junliang Li

> Thanks,
> 
> Carlos Konstanski
>