Just wanted to know, after a switchover, how were you able to get back to original state with out copying data to the source from distinate ? Can you put your steps in here ?
I had to allow replication to occur from the TARGET back to the SOURCE. This was something I was hoping not to have to do, but it makes sense if you think about it.
Because this is a controlled switch over and you want the resources to come online, (even though this was just for a test) in order to Switch BACK over, the VDMS, Replication sessions and the File Systems all have to be in a specific state.
From the Target (now the Source) started replications of the VDM and File Systems (using the GUI to perform these functions).
Once I verified the replication was from Target back to Source, I could now perform the "SWITCH OVER" again.
I did that and all was right with the world.
So, to recap,
I WANTED to perform the switch over without repicating back to the Source from the Target. Since the Target becomes the Source during the switch over, you must allow for that to occur before you can switch back.
My manager has asked for this test again, so I hope to be able to provide additional information. I dont have an ETA on when this will occur, but once it does, I will edit this with specific CLI commands.
Thanks for the information. So is there a way we can discard the test data on the target and get back to normal setup replicating from source to target back or NOT ?
In my opinion, it is not considered "TEST" data, unless you specifically put some files/folders what have you, in a specific location and then delete them after your testing is complete.
Because of the nature of the switch over, the target becomes the source until such time as you switch it back.
Who knows who may be modifying a file, adding data, etc while the resources are running on a different Celerra.
THe only other thing to do, would be create a TEST VDM, CIFS servers and File system and play with it that way. (in hindsight, this is what I should have done).
We have performed a switch over test again and am leaving it switched over for an extended period of time.
While examining the target celerra (where the switched over VDM and FS are currently running) we noticed that when viewing the CIFS SHARES, there arent any listed in Celerra manager. Is this by design? If not, is there an issue? Can we create new shares while "switched over" and if so, can that cause any issues with the existing shares that were created while running on the source?
We are migrating a CIFS server (VDM and FS) from NAS02 to NAS03 with Celerra Replicator v2.
After follow the steps to switchover VDM and FS, we noticed that in destination (NAS03), the IP in destination did not respond to ping.
We found a problem with the Cisco Switches and VLANs..... the VDM and FS now are in a switched over state (Source is in RO and destination is in R/W). As we found a problem, we cannot continue with the migration and we have to roll back, there is no sense to reverse the replication, I'ld like to resume the replication from source (NAS02) to destination (NAS03) discarding the changes in NAS03 (there was no changes because the IP doesn't work...)
So In this way there must be a way to recover the original replication from a switched over state....
I am not so sure about the VDM but for the FS, make the target FS RO and Source as RW and resume the replication discarding the changes on the target FS
DHoffman2
2 Intern
•
306 Posts
0
April 7th, 2010 12:00
Well, I answered my own question....
If anyone has any questions on my procedure (which was modified from what I posted) let me know.
Manumari
131 Posts
0
April 9th, 2010 12:00
Hi,
Just wanted to know, after a switchover, how were you able to get back to original state with out copying data to the source from distinate ? Can you put your steps in here ?
Thanks,,
DHoffman2
2 Intern
•
306 Posts
0
April 13th, 2010 10:00
I had to allow replication to occur from the TARGET back to the SOURCE. This was something I was hoping not to have to do, but it makes sense if you think about it.
Because this is a controlled switch over and you want the resources to come online, (even though this was just for a test) in order to Switch BACK over, the VDMS, Replication sessions and the File Systems all have to be in a specific state.
From the Target (now the Source) started replications of the VDM and File Systems (using the GUI to perform these functions).
Once I verified the replication was from Target back to Source, I could now perform the "SWITCH OVER" again.
I did that and all was right with the world.
So, to recap,
I WANTED to perform the switch over without repicating back to the Source from the Target. Since the Target becomes the Source during the switch over, you must allow for that to occur before you can switch back.
My manager has asked for this test again, so I hope to be able to provide additional information. I dont have an ETA on when this will occur, but once it does, I will edit this with specific CLI commands.
Manumari
131 Posts
0
April 13th, 2010 11:00
Thanks for the information. So is there a way we can discard the test data on the target and get back to normal setup replicating from source to target back or NOT ?
thanks
DHoffman2
2 Intern
•
306 Posts
0
April 13th, 2010 11:00
In my opinion, it is not considered "TEST" data, unless you specifically put some files/folders what have you, in a specific location and then delete them after your testing is complete.
Because of the nature of the switch over, the target becomes the source until such time as you switch it back.
Who knows who may be modifying a file, adding data, etc while the resources are running on a different Celerra.
THe only other thing to do, would be create a TEST VDM, CIFS servers and File system and play with it that way. (in hindsight, this is what I should have done).
Manumari
131 Posts
0
April 13th, 2010 11:00
Good to know. Helped a lot. Thanks very much for the input and thanks everyone as well.
umichklewis_ac7b91
300 Posts
0
April 13th, 2010 11:00
Excellent! When you post your commands, can you ask the mods to make this a sticky/permanent entry? It would be extremely handy.
Thanks!
Karl
DHoffman2
2 Intern
•
306 Posts
0
July 21st, 2010 06:00
We have performed a switch over test again and am leaving it switched over for an extended period of time.
While examining the target celerra (where the switched over VDM and FS are currently running) we noticed that when viewing the CIFS SHARES, there arent any listed in Celerra manager. Is this by design? If not, is there an issue? Can we create new shares while "switched over" and if so, can that cause any issues with the existing shares that were created while running on the source?
Dr.diSousa
1 Rookie
•
97 Posts
0
December 21st, 2011 16:00
Hi, this was my situation...
We are migrating a CIFS server (VDM and FS) from NAS02 to NAS03 with Celerra Replicator v2.
After follow the steps to switchover VDM and FS, we noticed that in destination (NAS03), the IP in destination did not respond to ping.
We found a problem with the Cisco Switches and VLANs..... the VDM and FS now are in a switched over state (Source is in RO and destination is in R/W). As we found a problem, we cannot continue with the migration and we have to roll back, there is no sense to reverse the replication, I'ld like to resume the replication from source (NAS02) to destination (NAS03) discarding the changes in NAS03 (there was no changes because the IP doesn't work...)
So In this way there must be a way to recover the original replication from a switched over state....
Any idea how to do it?
Manumari
131 Posts
0
December 21st, 2011 18:00
I am not so sure about the VDM but for the FS, make the target FS RO and Source as RW and resume the replication discarding the changes on the target FS
dynamox
9 Legend
•
20.4K Posts
0
December 21st, 2011 21:00
you will need to start replication in reverse order and then reverse replication
nas_replicate -start repl_fs01 -reverse
nas_replicate -reverse repl_fs01
julianojeziorny
2 Posts
0
September 10th, 2015 05:00
Not entirely true, one can in fact restart a switched over replication without hav ing to reverse it first by making the destination volume RO first:
On destination:
server_mount server_2 -option ro FS_NAME
Then on source start the replication with overwrite_destination option
nas_replicate -start REP_NAME -overwrite_destination