[OpenAFS-devel] advisory locks on RO vols

[Derrick Brashear]

I don't think there's any problem here, at least I have seen none to this point.

On Tue, Oct 2, 2012 at 5:54 PM, Andrew Deason <adeason@sinenomine.net> wrote:
> Gerrit 8197 makes the unix client effectively ignore advisory lock
> requests for files in RO volumes. I'm pretty sure this is okay, but I
> just wanted to raise it here to give people more of a chance if there's
> any possible issues with that.
>
> This should be okay, since for an RO vol, there's no writes that we can
> conflict with from another process. There should be no write locks we
> can conflict with, since a process cannot issue a write lock for a file
> opened read-only, and you can only open files read-only in an RO vol.
>
> And the fileserver already does not track locks for RO vols, which makes
> me think I'm not missing anything. At some point I thought maybe it
> would be possible for processes to misuse locks as some kind of general
> IPC mech or something, which ignoring them would disrupt (whereas right
> now they are tracked in-kernel even if not on the fileserver), but...
> I'm not even sure that's detectable by an application. Since, afaik you
> can only ask if there is a conflicting lock held, not just "are there
> read locks" or something.
>
> --
> Andrew Deason
> adeason@sinenomine.net
>
> _______________________________________________
> OpenAFS-devel mailing list
> OpenAFS-devel@openafs.org
> https://lists.openafs.org/mailman/listinfo/openafs-devel
>



-- 
Derrick