Doing a local replication of 4TB file system fs1(R/W) to fs1_replica1(R/O) and since the source keeps on changing I want to force source fs1 to R/O mount and do a manual sync and then do cutover to target.
Can I do force R/O on production file system ? Will it propagates R/O settings to all data just like permission change OR it applies "ro" to just the mountpoint and hence all data underneath it automatically become R/O ? I just want to make sure it does not propagate R/O settings to each and every files in 4TB.
There's no "propagation." The filesystem is set to RO not the files themselves. Switchover is still applicable even though the fs is on a physical dm and not on vdm. When you perform a Switchover the fs on the target side becomes RW. The only reason you want to set the source fs to RO is to guarantee consistency, correct? That's fine and there are no downsides to setting the source fs to RO.
The local replication target (fs_replica1) is on a physical DM instead of vdm. So failover, reverse will not help on this I believe. But if I set r/o attributes to filesystem will the r/o tag propagates thoughout the filesystem of 4TB just like the permission change? If it happens I believe it will take hours to complete. Any ideas?
server_mount server_2 -option ro fs1 /vdmx/fs1 -> this will set the whole filesystem to read only. Anything that you have on that filesystem will become read only.
I recommend you use the Celerra Replicator failover options like Rainer suggested. Use "switchover" once you set the filesystem to read-only and performed a sync.
I would suggest to revisit the Replicator and VDM manuals
Replication itself never changes mount points or shares
Of course if you are replicating a VDM then the VDM will carry the CIFS shares and replicate / activate them. Their path doesn't change since it's relative to the VDM root
For NFS you would need to change exports manually
Or just mount the fs onto the old mount point (not really necessary in a CIFS only VDM config)
If I'm understanding your question you want to change the mountpoint on the target fs from the default mountpoint to something different, is that correct?
When you create a fs it will mount it using default settings (fs1 on vdm1 will be mounted as /vdm1/fs1). If you want to change that (from /vdm1/fs1 to /vdm1/fs1_new for example) you have to delete the mountpoint and recreate it. You can do that within Celerra Manager.
Since the source fs is not on a vdm permissions/shares will not be migrated. You have to use a tool such as sharedup to accomplish that.
Thanks Ernes. I just did a ro mount of a testfilesystem and changed it back to rw and it works. Yes, I want consistency for the target FS.
If I do a switchover will the NAS DB get modified with new filessytem path for the same share? I'm talking abt this file " cat /nas/server/vdm/vdm_x/export.shares"
I want to mount fs_replica1 on same path as /vdm_x/fs1 so that there is no change in permissions etc.
Will validate ro option on production tonight. Thanks for the help.
Agreed. Replicator will not change the mountpoint. However, you can change the mountpoint manually. If the target fs is created by replicator then it will use the default mountpoint. You can go to Filesystems -> Mounts and delete the mountpoint and re-create one. If replicator created a mountpoint for fs1 to be /vdm1/fs you can go into Filesystems -> Mounts, delete the /vdm1/fs1 mount and recreate one (i.e. /vdm1/fs1_new)
etaljic81
1000 Posts
725
0
Posted July 8th, 2011 10:00
There's no "propagation." The filesystem is set to RO not the files themselves. Switchover is still applicable even though the fs is on a physical dm and not on vdm. When you perform a Switchover the fs on the target side becomes RW. The only reason you want to set the source fs to RO is to guarantee consistency, correct? That's fine and there are no downsides to setting the source fs to RO.