dual vio paths: hcheck_interval disk attribute 1

[hfaix]

Does anybody know if a reboot is required after changing the hcheck_interval attribute of a mpio client hdisk (of a dual vio config)? This allows for automatic path recovery if a vio goes down and comes back up.

The APV redbook (sg247940) doesn't say anything about it, but I found this link that says it's required.

http://techsupport.services.ibm.com...n/configuring_mpio_for_the_virtual_client.pdf

It would really suck if a reboot was required everytime a disk was added, because by default I believe health check is off.

 

[p5wizard]

If you turn the healthcheck interval on before the disk is used (put into an active VG), then no reboot is necessary.

If the disk is already active, you can only modify the parameter in the database (chdev ... -P) and a reboot is necessary for the attribute to become effective.

Other option: unmount, varyoffvg, chdev (without -P), varyonvg, mount.

If you have test VIO client, you can try it out for yourself. Remember to check your other VIO clients for inactive paths and reactivate them manually if necessary.

Hint: you don't even need to stop/start the VIO server for this test, simply remove a vtscsi device for the test client from one of the VIO servers, look at the VIO client to see what happens and then rebuild the vtscsi device and see it become active again.

But a real world test would include VIO server shutdown (or even: halt w/o shutdown) and then reboot of the VIO server.



HTH,

p5wizard

 

[hfaix]

Great post. Thanks