Solved!
Go to SolutionHow do I do a D/R test on my R2s (SRDF/A)?
All:
We have the following setup:
VMAX DMX
SRDF/A
R1s ------------> R2s -> GBCV
The SRDF/A runs continuously. All of our R2 devices would be "Consistent". When we do a D/R test we split the GBCVs and mount them on D/R servers at the D/R site and run our D/R test. Then when the D/R is over, we unmount the GBCVs, synch the R2s -> GBCVs and we're back to normal (by the way is the BCV sync full or incremental?).
Now our managers want to test running the D/R test on the R2s. So, I have mapped and masked the R2s to the D/R servers (HP-UX 11.23). So, what will the steps be:
1. Split the the R2s with symrdf -g rdfgXX split. This should bring the R2s to "Ready" status.
o Do I need to "suspend" the rdfgXX first?
o I don't want changes to "back up" in cache and fill up my cache and cause problems on the source VMAX.
2. Mount and run the D/R test on the R2s.
3. Three days later, we unmount the R2s.
4. Resync the R1s -> R2s. symrdf -g rdfgXX est
o Is this the right command?
o Is this an incremental establish or a full establish.
I don't want to do a "failover" or a "failback" I want Production to continue to run on the R1s undisturbed. Then I want to get back to normal as quickly and easily as I can with minimal impact to production and to the R2s (and GBCVs).
Stuart
2707
1
Posted January 28th, 2013 09:00
Hi Stuart again,
i am not so familiar with the SRDF/A like the SRDF/S but i will try to help you.
1. Split the R2 devices using the mentioned command is correct
a. It is better to suspend it
b. While suspend and especially SPLIT command will be executed no data remains on the cache slots.
2. Your devices are Ready and Write Enabled so you can mount them.
3. Unmount the R2 after 3 days
4. Establish the link between the R1 and the R2 from R1 to R2.
a. It is the right command
b. It is an incremental establish.
To perform a failover and failback you HAVE to EXECUTE the commands on your own!!!!!!
2707
0
Posted January 28th, 2013 12:00
I would suggest you add the -consistent option on your split command.
2707
1
Posted January 28th, 2013 14:00
before you start the split the SRDF/A session, issue the check point on the SRDF/A session.
1) suspend the SRDF/A session
2) issue the chekc point on SRDF/A symrdf -sid XXX -rdfg %% checkpoint (to make sure last cycle of data has been passed to the DR site)
3) split the SRDF/A session.
4) Map your R2 devices to Dr servers
5) once done unmount the DR volumes
6) Establish the R1-R2 link symrdf -sid xxx -rdfg %% est (this is an incremental est), pls note with this comamnds you loose all the changes that were made at the R2 side
Thanks
Sumanth
2719
0
Posted January 29th, 2013 10:00
All:
thanks for help. I have a question about commands and order of commands:
# symrdf -g rdfg20 suspend
# symrdf -g rdfg20 checkpoint
# symrdf -g rdfg20 split -consistent
#
# symrdf -g rdfg20 resume
# symrdf -g rdfg20 est
If you suspend, then you should resume. But resume and "est" are mutually exclusive, I think. Can you split a group that is suspended?
Stuart
2719
1
Posted January 29th, 2013 11:00
Stuart, yes the resume step will be replaced by EST.We donot have to resume a suspended session, just split it and establish it. plz find the sequence below.
ALso we can split a SRDF sesion which is in suspend mode.
IF MSC is enabled then we might have to use -force flag to split it .
RDF (RA) Group Number : 35 (22)
Source (R1) View Target (R2) View MODES STATES
-------------------------------- ----------------------- ----- ------ ------------
ST LI ST C S
Standard A N A o u
Logical Sym T R1 Inv R2 Inv K T R1 Inv R2 Inv n s RDF Pair
Device Dev E Tracks Tracks S Dev E Tracks Tracks MDAE s p STATE
-------------------------------- -- ----------------------- ----- ------ ------------
DEV001 08EE RW 0 12292 NR 0764 WD 0 0 A... X . Suspended
-bash-3.2$ sudo symrdf -cg srdf_app_test_cg split -force
Execute an RDF 'Split' operation for composite
group 'srdf_app_test_cg' (y/ ) ? y
An RDF 'Split' operation execution is
in progress for composite group 'srdf_app_test_cg'. Please wait...
Suspend RDF link(s) for device(s) in (0809,035)..................Started.
Suspend RDF link(s) for device(s) in (0809,035)..................Done.
Read/Write Enable device(s) in (0809,035) on SA at target (R2)...Done.
Read/Write Enable device(s) in (0809,035) on RA at target (R2)...Done.
Suspend RDF link(s) for device(s) in (0809,035)..................Started.
Suspend RDF link(s) for device(s) in (0809,035)..................Done.
The RDF 'Split' operation successfully executed for
composite group 'srdf_app_test_cg'.
-bash-3.2$ sudo symrdf -cg srdf_app_test_cg est
Execute an RDF 'Incremental Establish' operation for composite
group 'srdf_app_test_cg' (y/ ) ? y
An RDF 'Incremental Establish' operation execution is
in progress for composite group 'srdf_app_test_cg'. Please wait...
Write Disable device(s) in (0809,035) on RA at target (R2).......Done.
Suspend RDF link(s) for device(s) in (0809,035)..................Done.
Resume RDF link(s) for device(s) in (0809,035)...................Started.
Merge track tables between source and target in (0809,035).......Started.
Devices: 08EE-08F6 in (0809,035).................................Merged.
Merge track tables between source and target in (0809,035).......Done.
Resume RDF link(s) for device(s) in (0809,035)...................Done.
The RDF 'Incremental Establish' operation successfully initiated for
composite group 'srdf_app_test_cg'.
1127
0
Posted January 29th, 2013 12:00
If you still want the status to be split try putting it in ACP_disk and then split it..the status will be in split
2718
0
Posted January 29th, 2013 12:00
sumanthn:
First, how do you past text into Forum Window. I can't.
2nd, I can "split" my SRDF/A groups. They start as "Consistent", and I can "Suspend" the RDFG or not. It doesn't seem to matter. The question is: Note that my R2s are "Failed Over" and not "Split". I don't want the R2s"Failed Over" - I want them "Split".
Then after I "split" them, I can re-establish them. But I'm worried about the "Failed Over"....
2718
0
Posted January 30th, 2013 18:00
All,
Its showing failed over because either of the devices (source or target) is not mapped to the front end. This is normal.
If a device is not mapped to the front end, split state would always show failed over.
regards,
Saurabh Rohilla
2005
0
Posted January 30th, 2013 19:00
Saurabh, I don't think mapping devices would change the status from failed over to split.
Sumanth
sumanthn
20 Posts
2719
0
Posted January 29th, 2013 12:00
Usually Async session when you split the status is failed over.you souldn't worry much as both the R1 and R2 are RW .
The status is shown as split in case of sync replication.
You should be good with failed over status as your applications can run both on R1 and R2 ,
I directly copy paste the text from my screen.