Re: DMAPI unmount bug?

[Dean Roehrich <roehrich@sgi.com>]

>From:  "James A Goodwin" <jagoodwi@us.ibm.com>
>All,
>
>I'm having problems with XFS DMAPI when an unmount fails.  My program is
>accessing the file system and receives a preumount event.  It tells DMAPI
>to continue, but was unable to relinquish all open files in the file system
>for some reason or other.

The HSM is supposed to take this opportunity to "close any active accesses . .
. so that [the HSM] application itself does not cause an unmount to fail."

You're responding to the preunmount too early, or you're responding with
"continue" when you know it's not okay to continue.


>  This causes the unmount to fail, and the DMAPI
>unmount event that is received via DMAPI shows this.
>
>However, the root user who issued the unmount command sees a successful
>unmount.  Subsequent runs of the 'mount' command show that the file system
>is unmounted.
>
>So there is a disparity between what is reported via DMAPI (unmount
>failure) and what is reported to the user (success). In this case, I'm
>guessing it really does unmount.  If that is true, then the DMAPI should be
>told that the unmount was successful.

There may very well be a bug here.  I'll look at it as soon as I can get the
dmapi module to build again...

Does /proc/mounts agree with /etc/mtab?

Dean