monitoring HBA link status on Red Hat linux

[vijay alur alur]

Hello All,

How do i monitor the HBA link status on my server running Red Hat Linux OS. On some instances, when there is a link down our storage team notifies us about one of the link to the LUN is down. There is no monitoring from our side that will generate a ticket to us. I am wondering how to monitor the HBA link down status.

Like for e.g. on HPUX would show error message in the file /var/adm/syslog/syslog.log. The same does not happens in case of Red hat linux.

Thanks,

VJ

[Matti_Kurkela · Accepted solution]

On RHEL 5 and newer, you can find the HBA link state at /sys/class/fc_host/host<number>/port_state.

You did not specify what kind of monitoring system you're using, but it should usually be easy to make your monitoring system check that too.

The port_state pseudo-file might be available on RHEL 4 too, depending on the HBA driver manufacturer and driver version.

In RHEL 4, the /sys/class/fc_host sub-tree was very new and not all drivers implemented it the same way.

For more background information, please see this thread:

http://h30499.www3.hp.com/t5/System-Administration/sys-class-fc-host-entries-missing-on-a-few-RHEL-4-and-5-systems/td-p/4827915

[vijay alur alur]

Hi Matti,

Thanks for the reply.

We have Red Hat Linux 5.7's and some 6.3's...

We use nimsoft and nagios to monitor and ticket gets generated.

I did come across the file you metioned /sys/class/fc_host/host0/port_state which showed Link Down. i was planning to write a cron job that would check this in a frequency. But i guess nimsoft can do this task much better. so monitoring issue should be fixed.

Also i have question about the HBA link going down. This has been happeneing for some time. now once we have monitoring set, i want to be able to reset the connection on the link down HBA so that it gets back online.

Currently whenever there is a one HBA link down, we have to ask Storage team to reset the connection to fix. We would want this to be fixed from host/OS side.

Thanks,

VJ

[vijay alur alur]

Accepted this post as a solution and kudos given. Thanks Matti for your help always!!!