Unsolved
This post is more than 5 years old
44 Posts
0
4003
Dynamic concurrent SRDF failover/failback scenario?
Hi ,
I have few questions regarding my Concurrent SRDF failover/failback scenario.
Let me explain my setup first. I have three sites i.e. Site A (DMX4-6 ,Mcode 5773) , Site B(DMX800 , Mcode 5671) and Site C (VMAX-1 , Mcode 5874). We have SYMCLI v7.3 installed on all three site nodes. We have dynamic Concurrent SRDF configured in the following manner:
Site A (R1)-------sync replication with---------------- Site B (R2). and
Site A (R1)---------async replication with-------------------- Site C (R2)
I have configured the concurrent SRDF setup properly and a host is connected to each site where SYMCLI v7.3 is installed.
Now we want to verify the failover scenario in case of host network failure, a complete site failure or two site failures. Let me start with step by step.
Scenario 1: Let’s say we have a production host on Site A and the host network fails(lost access to Site A storage) at site A( replication among all three sites are still working fine) and now we want to write the data from Site B.
So what commands or procedure we should run from Site B host( and/or Site C host). Please tell us in details.
Scenario 2: Let’s say we have Site A and Site B host network fail(Site A and Site B hosts have no access to Storage) and now we have one host that has access to storage i.e. Site C host. Now what commands should we run from Site C host to make R2 device R1 . Please explain in details.
Please let me know if I need to give more detail?
Thanks,
PedalHarder
465 Posts
1
July 18th, 2011 01:00
In both senarios the goal is to make the R2 devices read-writeable to allow the applications to come back on-line. You have device groups available at both remote sites?
It is either 'symrdf -g dg_name failover'
or
'symrdf -g dg_name rw_enable r2'
Provided it is not a complete disaster and your r1 array comes back online at some point, if you do a failover, then you can do a failback to get back to the R2 site from whatever R2 site you were running from. There will be a secondary step to re-synchronise the other concurrent SRDF leg after the failback.
If there is no access to the R1 array then any command that needs to talk to both arrays (eg symrdf swap) will no execute.
Shubh_Ram
44 Posts
0
July 18th, 2011 03:00
HI Jasonc,
Yes we have Device Group created on all the sites. Let's say it is not a complete disaster but we still want to swap the roles and want to make secondary device as R1 as per both scenario described above.
Thanks,
Shubh.
vijay861
51 Posts
1
July 18th, 2011 06:00
Best Practice for your Environment was,
Create Device groups with same names on all the three Sites. Source as RDF1 type and Targets as RDF2.
command is "symdg -type RDFXXX create XXXX_name.
In case of site failure(As mentioned in scenario 1 and 2):- if you want to change R2 devices as primary. Then you can swap the Devices in the device group.
command is "symrdf -g XXXX_name swap ( if you want there is an option that you can swap only specified devices also).
In case of failover, (As discussed by other participant, This is temporary Disaster. So we can use failback in case R1 devices are ready to use.
command " symrdf -g XXXX_name failover.
The above mentioned commands are supposed to run on SITE where there is active connection to host.
Shubh_Ram
44 Posts
0
July 18th, 2011 22:00
Thanks Venkat,
Ok let me explain what I want to do.
In scenario 1 , where I want to make Site B as R1 devices . For that we have to run , symrdf -g xxx failover -rdfg 00 then swap but when you run failover then Site C devices come in Invalid state. and we have to run "rw_enable r1"on these devices. then they come in consistent state(because Site C is in Async mode with Site A). Now we can swap roles on Site B devices and can resume suspended state. Now Site B devices become R1. This is happening very well.
But when I try to do same with Site C(which is in Async with Site A), It is not happening. When we try to run swap roles on Site C, devices come in "Split" state and if we establish(or restore remote) then again they become R2.
So the problem is with Async mode.
Any suggestion would be appreciated.
Thanks in advance.
Shubh_Ram
44 Posts
0
July 19th, 2011 23:00
Can someone help me out in this, or point me out some good docs for this ?
Thanks,
vijay861
51 Posts
1
July 20th, 2011 08:00
Hi Shubh_Ram,
you can have a glance at this document.
which may helpful for you
https://powerlink.emc.com/nsepn/webapps/btg548664833igtcuup4826/km/live1/en_US/Offering_Technical/Technical_Documentation/300-001-165.pdf?mtcs=ZXZlbnRUeXBlPUttQ2xpY2tDb250ZW50RXZlbnQsZG9jdW1lbnRJZD0wOTAxNDA2NjgwNWMwNzkxLGRvY3VtZW50VHlwZT1wZGYsbmF2ZU5vZGU9MGIwMTQwNjY4MDUxNzBmY19Hcmlk
Shubh_Ram
44 Posts
0
July 21st, 2011 23:00
Thanks Venkat for the docs. Will check if I find something which I needed.
Thnaks,
Shubh
vijay861
51 Posts
0
July 22nd, 2011 06:00
If these answers are helpful then mark your question as Answered.
It might be helpful for the people with same questions.
Thanks & Regards,
Venkat
SKT2
2 Intern
2 Intern
•
1.3K Posts
0
August 26th, 2011 07:00
when you did a fail over between (siteA and siteB), the SiteB devices became RW and siteA device became RD only and siteC device became invalid. Once siteA is RW_enable state srdf/A to SiteC became consistent. Then you do swap and siteB become R1 and siteA became R2. At this time what is the status of the siteA devices??(WD or RW enabled)
how can you do "swap" at this time between siteC and siteA(when siteA is no longer R1).? In my view , your options may be go back to siteA when its available back again or setup a new srdf/a between siteB and siteC if the recovery of the siteA takes too much time.