Unsolved
This post is more than 5 years old
4 Posts
0
1929
September 26th, 2018 16:00
failover and set_production_copy commands
I had a couple questions about failover with a consistency group with more than one non-production copy (using the CLI) if anyone could help:
1. Is the failover command synchronous? In other words, will set_production_copy work if say the failover command is done, then the set_production_copy command is done immediately afterwards (as in a script)?
2. The set_production_copy command appears to have a limitation. The CLI manual says "Before defining the remote copy as the production copy, ensure there is only one copy at the original local cluster". What, exactly, is the "original local cluster" in this context? Does this even matter (if the "remote copy" in this context has become the previous production copy)?
Thanks for any help anyone can provide!
forshr
2 Intern
•
1.1K Posts
0
September 27th, 2018 06:00
Set Production is only relevant with multi-copy replication, i.e more than one replica copy.
If you failover with multi-copy CG, you will be given options relating to the non-Production replica copy(ies) not including the original Production copy.
Regards,
Rich
mcdonomr
4 Posts
0
September 27th, 2018 10:00
That hasn't been the case (no options are given) when I've done, say:
set_production_copy group= copy= -f
to a group with more than one copy, rather than using the graphical interface. Perhaps default values are assumed when you do that? There are other parameters listed in the manual that I don't think I need to use.
Additionally, the manual says:
"Before defining the remote copy as the production copy, ensure there is only one copy at the original local cluster."
I don't know what this means. Does that mean you can't do set_production_copy in certain situations?
Thanks for the help!
Matt
forshr
2 Intern
•
1.1K Posts
0
September 28th, 2018 12:00
Hi Matt,
The -f will force the full sync to the additional replica(s).
The comment from the manual is wrong and doesn’t make sense.
Rgds,
Rich
mcdonomr
4 Posts
0
October 8th, 2018 10:00
Rich,
Thanks for the help. I had an additional question if you are able to answer: When doing set_production_copy through the CLI (using more than one non-production copy - i.e. a production, local, and remote copy where I did a failover from the production copy to the remote copy), I have in the past had it work without error in a test environment, however I have also seen it come back with:
“Error: Operation failed. Each enabled target must have a link to the production copy. (Create the following link: ORIGINAL_REMOTE_COPY->ORIGINAL_LOCAL_COPY before setting production)”
Note: the instance in which I have seen it work without this error was using async replication in a test environment.
So my questions are this:
1. Why does this happen (and why does it happen apparently only sometimes - perhaps it only happens the first time or it has something to do with the protection policy)? Does RP not establish links on it's own (except to the previous production copy)?
2. I assume this CLI error/behavior is analogous to the Set Production dialog box described in the RP Administrator's Guide (Under Recovery->Manage Recovery->Failing over, also seen here: https://www.youtube.com/watch?v=SrHCpuoTsA4#t=06m17s). Is that correct?
3. The proper response in this scenario would indeed be to do the create_link, then retry the set_production_copy, correct? Are there any side effects of creating a new link that I need to be worried about?
Thanks,
Matt
forshr
2 Intern
•
1.1K Posts
0
October 9th, 2018 01:00
Hi Matt,
It shouldn’t happen and I would recommend raising and SR with Support.
Regards,
Rich