HPE Storage Users Group

Hi all,

a 3par System with 4 8450 Subsystems.
sub01 is remote copying to sub02
sub03 is remote copying to sub04

For example:
I create a VV on sub01 and Export it to a Server
I create a remote copy on sub02 and Export it too

Is it correct to Export it? Makes sens in my opinion if sub01 Fails.
The Problem is that it takes a lot of time when the remote copy is exported and i rescan the disks in disk Management on the Server (about 10-15 seconds). I get disk latencies up to 2,5s and i think it has sth to do with the RCP (but i have to test more).

So the question is: do i Export the rcp?

Thanks

Daniel

If you are using Peer Persistence, yes.(remember path_management, if not you are trying to write to a read-only volume)
If you are using CLX, you export it to another host
Otherwise, no.

_________________
The views and opinions expressed are my own and do not necessarily reflect those of my current or previous employers.

Only if your setup is configured and supports the 3PAR Peer Persistence feature otherwise not a good idea.

Check the HPE 3PAR Remote Copy guide and the 3PAR Host OS Implementation Guides for your server.

It needs certain versions of things and settings at the array and server ends and if you want it to auto failover between arrays you'll need a Witness server setup at a 3rd 'site'.

thank you guys,

normally our System design is made for peer persistence with a Quorum Server. When i check the Quorum state, it hangs on "re-starting". So there has to be a Problem says the documentation.

The main Goal would be:
10 VV exported to three Hosts (SQL Server)
remote copied to other Sub System

Cluster controlled by Microsoft Server Cluster and MPIO.

Should work, right?

Oh, witness check says: no route to Quorum witness

So it could be a big Problem, when i Export from origin and remote copy the "same" volume when the Quorum is not active? Or is this a Thing that MPIO can handle!?

The Quorum/Witness is only used for auto failover and only if the Rcopy Links between arrays go down, otherwise it should be totally passive as far as LUN access is concerned.

MPIO on the servers needs configuring correctly for Windows I think, correct driver/firmware installed. Rcopy group policy needs at least path_management set, host objects for windows I think needs persona 15. Preferably both copies of the LUN need same LUN ID and WWN. The guides will list how to check the MPIO path statuses at the hosts (copy paths should be 'standby').

There's a Witness connection test command too, maybe some ACL/firewall issues if not reachable and up and working/correct version etc.

Okay, how can i see the host persona? Win2012 is in the drop down, but does that really Count?

Ok, so our main Problem:

Volume exported to SQL1 and SQL2 (same volume)
Copy exported to SQL1 and SQL2

in MPIO i see 4 Active and 4 Standby paths. Thats good i think.
On SQL1 the disk is active, on SQL2 it Shows reserved.

When we perform a Cluster Switch from SQL1 Server to SQL2 server, you should see the disks go offline on SQL1 and go online on SQL2 within a few seconds.
Sometimes, it really runs into a timeout after 15 MINUTES..only the disk Change!! And i don´t have a clue where to look anymore. HPE says, everything is finde on their end/system, so not very helpful

It worked a few months ago, but of course no one knows of any changes made. The amount of disks (at least 30 exported) could be a matter?

The text field and the setting used to be completely separate but the current GUI tries to set both correctly when you pick the OS now.

For Rcopy it's worth getting used to the CLI commands as there are bits still a little messy in the GUI.


Persona_Id should be 15 on both arrays, in the last column it lists ALUA, this is the scsi feature that is actually behind the Peer Persistence stuff.


The last column (ID) is the 'Target Portal Group ID' or TPGID that the array is sending to the server.

1 = normal LUN
257 = source/active PP LUN
258 = copy/standby PP LUN


The group containing the VVs should be started and have at least "path_management" in the Options.

I've only got one test Windows box (2012R2) as all our hosts are VMware/Solaris so this bit is just what is working here.

In Computer Management/Disk Management look at the properties of the 3PAR LUN and go to the MPIO tab, Policy at the top think should be set to 'Round Robin With Subset'. In the Paths section you should see the 'active' paths from the source array (again with TPGID 257) and the 'Standby' paths from the copy array (TPGID 258).

It should all be working if that's correct.

HBA drivers/firmware, Windows hotfixes, 3PAR OS version/patches can cause issues if not at supported levels, the guides and HPE SPOCK site should cover all that.

Checked for one disk

3 sq3 15 WindowsServer UARepLun,SESLun,ALUA,WSC


9 sq2_dblog sq3 2001000E1EE970F2 2:0:2 host active 257

9 sq2_dblog.r sq3 2001000E1EE97230 3:0:1 host standby 258

The RC Group has path_management enabled. And i can confirm the the same TPG is listed in the MPIO paths.