Red Hat Advanced Server 2.1 cluster

[Troels Rasmussen]

Is there anyone who ever have seen a RHAS2.1 cluster, which worked?

Brgds Troels

[Schelstraete Bart_1]

Do you mean the cluster provided by RedHat, or the HP one? (serviceguard?)


Bart

[Troels Rasmussen]

The one provided by Red Hat !

[Steven E. Protter]

I've seen them working, but don't own or use the product.

If you search for Red Hat Advanced Server 2.1 cluster in the search option above on this page, you'll see lots of threads and questions and answers on the topic.

Properly configured my itrc colleagues say it works quite well.

SEP

[Troels Rasmussen]

Not solving my problem ! HP Denmark said that they knew there was a problem with EVA SAN storage, but they couldn't tell me how it should be setup to work correctly? Maybe someone here knows?

Brgds Troels

[Vitaly Karasik_1]

as usually you should read HCL for RHEL + clustering; and unfortunately the list of certificied shared storage devices isn't long

Vitaly

[Troels Rasmussen]

Guess you are right, but has anyone heard of a RHAS 2.1 cluster solution with Red Hat's own cluster, which was working?

I have a problem making the quorum for the cluster which comes with RHAS21. It seems like it is not possible for the two nodes to share the SAN disk, one of the nodes likes to take control of the disk and then it is not possible for the other node to connect to it. It is possible for both nodes to see the SAN disk, but when one is up the other one can't... When you try to load the module on the other node which is not active, the first one looses access and the new node takes control.

I am using the qla2x00 driver which comes with the kernel 2.4.9-e.25 (qla2300) with a qla2312 card, this is compiled as a module for the servers.

I get an error like this (dmesg),

Device 08:00 not ready.
I/O error: dev 08:00, sector 0
Device 08:00 not ready.
I/O error: dev 08:00, sector 0
Device 08:00 not ready.
I/O error: dev 08:00, sector 0

We are using storageworks EVA 8000 with fabric SAN switches!

There is an IBM doc at http://www-1.ibm.com/servers/esdd/articles/redhat/ which describes how it should be setup with IBM equipment, what about HP?


Brgds Troels

[Vitaly Karasik_1]

I built quorum partitions without problem on shared SCSI storage connected to two DL380G3
(SmartArray5 controller, as far as I remember)

Vitaly

[Virgil Chereches_2]

What kind of setup do you have?
I mean, the hosts are connected to storage through different controllers?
If that's the case you may need SecurePath on both hosts.

[Troels Rasmussen]

Both servers has one qlogic adapter, connected to a storageworks SAN switch, which is connected to the SAN!

[MG_9]

Which storage do You use MA8000 or EVA. What is connection type for the LUN where Quorum is (which OS?).

[Troels Rasmussen]

Actually I think it is an EVA5000, it is brand new, contains some big disks in a raid set. I am not sure I understand the other question? It is connected by qlogic 2312 cards with fabric to a switch. OS is RHAS 2.1

[MG_GM]

It seems to me that You have problem with storage/SAN configuration. Check the following:
- Switch zones: SAN switch ports must be zoned so that both HBAs connect to only one port of one HSV controller (for instance port 1 on controller A). Make sure that both have path to the same port. This is because You do not have secure path to handle more than one path to the storage.
- Virtual disk (Your Quorum) on EVA should be presented to HSVA port1 as "Path A Failover/Failback). All other LUNs for the cluster must be presented the same way
- Host connections to EVA: Each Qlogic adapter has unique WWN. You need to configure access to EVA storage for each host. When adding hosts You need to set OS type to SUN Solaris!
On host side You must install EVA 5000 platform kit for Linux ( http://h18006.www1.hp.com/products/storageworks/softwaredrivers/enterprise/index.html ). It will update Your drivers and settings and create new ramdisk.

Let me know, how it goes.