This post is more than 5 years old
6 Posts
0
53370
Replication setup between PS5000X and PS6000XV
Hi.
Probably just some configuration I have missed, but I cant get any further.
We just bought a new PS6000 SAN to setup replication from a production PS5000 SAN. 3 VMware host uses to the PS5000.
I installed the PS6000 SAN following the manual and connected to the same switch as the ps5000 SAN. Then, on both SANS i added the other SAN as a replication partner using the Group name and the group IP address, and respective passwords.
SANPS5000(4 volumes)
Group Name: san1
Group IP: 192.168.1.1
Replication partner:
Group name: san2
Replication partner IP: 192.168.1.2
****************************************************************************
SANPS6000(no volumes added)
Group name: san2
Group IP: 192.168.1.2
Replication partner:
Group name: san1
Replication partner IP: 192.168.1.1
For some reason, when I try to setup replication on a volume on San1, I get a "No connection could be established. Verify that the partner IP address is correct" error message at the end.
I have verified that the IP's is correct, both SAN's are connected to only one switch(switch ports are unconfigured), the computer I use to setup replicaton is attached to the san switch and both group IP's can be pinged from this computer.
Firmware is 4.2 on ps5000 and 4.3 on ps6000
Any advice that might help me would be much appreciated.
\\Tor
Dev Mgr
4 Operator
4 Operator
•
9.3K Posts
1
September 17th, 2010 08:00
A few things:
- when defining the replication partner, the SAN doesn't 'test' if your information is right, so if you defined the partner incorrectly, you wouldn't know till you try to replicate an actual volume
- when defining a replication partner, the connection is case sensitive for the group name
- just to ensure the 2nd SAN is set up, you could add that group IP to the iSCSI initiator on one of your ESX boxes, create a dummy volume and verify that the ESX box can see the volume (don't need to format it or so, just verify that it's visible under the storage adapters)
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
1
September 20th, 2010 08:00
The 5000 has 3x GbE interfaces and the 6000 has 4, each physical port on the active CM needs an IP address in that subnet. The array uses connection load balancing to assign new iSCSI sessions to a unique physical interface. During the login process, the connection is redirected to a physical port on the array.
It sounds like only ETH0 on each array has been enabled. So you are severely limiting the I/O on that SAN. Plus you have no redundancy if that's true. A single cable/switch failure will bring SAN down.
FYI: Both arrays need to be upgraded to 4.3.7. There are some important fixes since 4.2.x.
Regards,
-don
jepp
6 Posts
0
September 20th, 2010 00:00
Thx a lot!
Case-sensitiv group name was the thing. But I cant understand why it is case-senstive and could not find this documented anywhere. I have been both googling and reading the documentation for the last couple of days without getting a hint about this.
One other thing, the documentation says that if you put a layer 3 address on each network modules interface, the network performance of the SAN would improve. As the technique used is Layer 2 multicasting I cant see how a layer 3 address on each interface would improve the network?
Any link to a network configuration document would have been nie:-)
\\Jepp
Dev Mgr
4 Operator
4 Operator
•
9.3K Posts
0
September 20th, 2010 07:00
Most people do replication between 2 different sites/buildings. When using a network that's that spread out, you don't want both locations to use the same subnet. So, if you are using 2 different subnets, you need a route between them so they can reach eachother. This is where the layer 3 address comes in.
If you have no intentions to move one unit to another site (especially if you're going to keep it in the same serverroom), why bother with replication? Unless you're just using it to learn about replication.
jepp
6 Posts
0
September 22nd, 2010 01:00
The SAN's will be on the same network using dedicated fibre cables between the sites/switches. I want the best performance, so if a layer2 mechanism is used within a SAN or replication between 2 SAN's, it seems for me to be the best option.
``Jepp
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
September 22nd, 2010 07:00
Hello,
I would agree with that. A flat L2 is preferred to a routed L3 configuration. One thing to note, is that at any one time only three replication tasks will be running at the same time. If you have more than that, it will timeshare across them. But no more than three will be actively transferring data at one time.
However, being on a local subnet should move the data quite quickly.
-don