Video Screencast Help
Symantec to Separate Into Two Focused, Industry-Leading Technology Companies. Learn more.

SF HA/DR With VVR Testing While Primary is Active

Created: 21 Oct 2012 • Updated: 25 Oct 2012 | 3 comments
yairz's picture
This issue has been solved. See solution.
 
Hi All,
 
I have a customer with SF HA/dr with VVR for Oracle (two servers in primary, one in secondary) on RHEL 5.
SF version is 5.1 SP1 RP2.
 
The customer would like to test DR site, without taking the primary site offline.
I cannot use Fire Drill due to various reasons.
 
What the customer would like is to dissconnect the network between sites, then bring the DR online, test the DB (connect an application to it) and then shut the DR site down and reconnect the replication lines.
All this should be done while the primary site is active.
 
From what I understand, as soon as I disconnect the line between the sites, I have to run a takeover on the DR site to enable data access and mount the volumes, but this will result in an primary-primary configuration error once I'll reconnect the sites.
Usually this is resolved by taking the original primary (now acting secondary) and forcing it to be a secondary, but I obviously cannot do that since the primary has to remain active at all times and will be more updated when I'll finish testing the DR.
 
So, is there a way to perform this task without destroying all the configuration when done or without leaving the DR site as a primary?
 
Ideally, I would like to perform the following:
1. Disconnect the links (Planning to do so by taking IP resource offline)
2. Mount the volumes at the DR site
3. Start the DB at DR site and test it
4. Stop the DB and unmount volumes
5. reconnect the links and mark the DR site as secondary.
 
Help would be much appreciated.
Thanks,
Yair 

 

 

Comments 3 CommentsJump to latest comment

mikebounds's picture

Firstly, can you not over come the reasons you cannot do a Firedrill as this is the best solution - only reasons I can think of is:

  1. Not enough of storage - only need generally about 10%, depending on how long snapshot is in place and how many changes are been made at Primary and on snapshot
     
  2. Enterprise licence is required do a space-optimised snapshot. 

Are these the reasons?

You can do what you want to do, however, after testing DR site, you must make DR a secondary (run vxrvg makesecondary) BEFORE you reconnect, otherwise the Primary will become an acting secondary.  You will have to do a full resync, so I would also detach rlinks before reconnecting network. Here is what I would do

 

  1. Freeze replication service groups at both sites and application service group at Prod
  2. Make Virtual replication IP resources non-critical at both sites
  3. Down Virtual replication IP using O/S tools at both sites (you won't be able to do in VCS because of dependencies)
  4. Verify that replication is disconnected
  5. Online application service group at DR, which will run vradmin takeover, Mount the volumes at the DR site and start the DB
  6. Test DR
  7. Offline application service group at DR which will stop DB and umount volumes
  8. Run "vxrink det" at both sites
  9. Run "vxrvg makesecondary" at DR site
  10. Unfreeze replication service groups and online Virtual replication IP resources 
  11. Make Virtual replication IP resources Critical at both sites
  12. Unfreeze application service group at Prod
  13. Run vradmin -a startrep at primary site to resync data

Mike

UK Symantec Consultant in VCS, GCO, SF, VVR, VxAT on Solaris, AIX, HP-ux, Linux & Windows

If this post has answered your question then please click on "Mark as solution" link below

SOLUTION
yairz's picture

Thanks Mike, this is a great step by step guide!

I just tried it in a Test environment and it works perfectly.

One additional note - I had to make the DB SG local in order to online it while the primary SG was online, ohterwise I would get an error.

I really appreciate your help with this issue.

 

Yair

mikebounds's picture

Ah yes, forget step to make global service group a local group for the duration of the test.

Mike

UK Symantec Consultant in VCS, GCO, SF, VVR, VxAT on Solaris, AIX, HP-ux, Linux & Windows

If this post has answered your question then please click on "Mark as solution" link below