DM manual failback doesn't work on RHEL5

DM manual failback doesn't work on RHEL5

[Yanqing_Liu]

[-- Attachment #1.1: Type: text/plain, Size: 1395 bytes --]

Hi all,

In our testing with DM RDAC stack with RHEL 5 (including 5.5 and 5.4), we found that manual failback doesn't work as expected.

What we did is to connect the host to Dell PowerVault 3200/i array. After everything is configured, we changed "failback" value for our device from "immediate" to "manual". Then the host is rebooted for the new value to take effect. However, automatic failback still happens once failed path is restored. It seems failback "manual" doesn't work.

We've also verified deferred failback works as expected. So "manual" failback seems to be the only issue.

Our device section in multipath.conf is as follows:

...
device {
                vendor                "DELL"
                product               "MD32xxi"
                path_grouping_policy  group_by_prio
                prio                  rdac
                polling_interval      5
                path_checker          rdac
                path-selector         "round-robin 0"
                hardware_handler      "1 rdac"
                failback              manual
                features              "2 pg_init_retries 50"
                no_path_retry         30
                rr_min_io             100
                prio_callout          "/sbin/mpath_prio_rdac /dev/%n"
        }
...

Any sugegstion or insight on this issue?

Thanks,
Yanqing


[-- Attachment #1.2: Type: text/html, Size: 4420 bytes --]

[-- Attachment #2: Type: text/plain, Size: 0 bytes --]

Re: DM manual failback doesn't work on RHEL5

[Malahal Naineni]

Yanqing_Liu@Dell.com [Yanqing_Liu@Dell.com] wrote:
>    Hi all,
>     
>    In our testing with DM RDAC stack with RHEL 5 (including 5.5 and 5.4), we
>    found that manual failback doesn't work as expected.
>     
>    What we did is to connect the host to Dell PowerVault 3200/i array. After
>    everything is configured, we changed "failback" value for our device from
>    "immediate" to "manual". Then the host is rebooted for the new value to
>    take effect. However, automatic failback still happens once failed path is
>    restored. It seems failback "manual" doesn't work.

There are few cases where you need to re-build your initrd if you change
multipath.conf file. What does the loaded table say? Does it indicate
that the failback mechanism is 'manual'?

Thanks, Malahal.