dwdha at drdykstra
Sep 8, 2009, 1:06 PM
Post #4 of 5
(1035 views)
Permalink
|
You shouldn't have to set the fsid if you're using DRBD properly.
I think the DRBD/NFS page is more out of date than the "See Also"
link there, http://linux-ha.org/HaNFS. See especially the "Device
Numbering" section.
I don't know how the use of reiserfs might affect this, I have no
experience with that, only ext3.
- Dave
On Tue, Sep 08, 2009 at 01:53:43PM -0400, Michael Kershaw wrote:
> when you export the file systems, you need to give them a consistent identifier:
>
> /mounts/homes/home1 *(sync,rw,no_root_squash,fsid=1)
> /mounts/homes/home2 *(sync,rw,no_root_squash,fsid=2)
> /mounts/homes/home3 *(sync,rw,no_root_squash,fsid=3)
> /mounts/homes/home4 *(sync,rw,no_root_squash,fsid=4)
> /mounts/homes/home5 *(sync,rw,no_root_squash,fsid=5)
> /mounts/homes/home6 *(sync,rw,no_root_squash,fsid=6)
> /mounts/homes/home7 *(sync,rw,no_root_squash,fsid=7)
>
>
> the nfslock contains the major:minor #'s unless you specify otherwise.
> if the major:minor of the device is different on the partner
> node(extremely possible as their dynamically assigned) the client
> won't be able to write to it. as long as you assign it something
> unique like above, you should be good to go!
>
>
>
> good luck!
> mike
>
>
>
>
> On Tue, Sep 8, 2009 at 11:51 AM, Dejan Muhamedagic<dejanmm [at] fastmail> wrote:
> > Hi,
> >
> > On Tue, Sep 08, 2009 at 04:45:53PM +0200, Harald Dunkel wrote:
> >> Hi folks,
> >>
> >> I am setting up a Linux NFServer together with drbd and Heartbeat 2
> >> (old format) on Lenny.
> >>
> >> ? ? ? kernel is 2.6.29.6
> >> ? ? ? drbd is 8.3.2
> >> ? ? ? heartbeat is 2.1.3-6lenny1
> >>
> >> For testing I am running 15 kernel builds on this NFS
> >> partition in parallel in an endless loop. Every 10 minutes
> >> the HA hosts change their role (using /etc/init.d/heartbeat stop).
> >>
> >> If the primary heartbeat is stopped (or if I run hb_standby), then
> >> I get a "Stale NFS file handle" message on all writing NFS clients
> >> after 3 minutes. 2 minutes later the clients start writing on the new
> >> primary :-(. First I thought that is what you deserve if you do NFS,
> >> but according to
> >>
> >> ? ? ? http://www.linux-ha.org/DRBD/NFS
> >>
> >> it seems that this is not supposed to happen. Is this correct?
> >> The haresources file is
> >>
> >> nasl002a IPaddr::xx.xx.xx.xx/23/bond0 \
> >> ? ? ? drbddisk:: \
> >> ? ? ? Filesystem::/dev/drbd1::/common::reiserfs::defaults \
> >> ? ? ? Filesystem::/dev/drbd2::/space::reiserfs::defaults,noatime \
> >> ? ? ? nfs-common \
> >> ? ? ? nfs-kernel-server
> >>
> >> The link mentioned above suggests to use something like
> >>
> >> nasl002a ? ? ?drbddisk:: \
> >> ? ? ? ? ? ? ? Filesystem::/dev/drbd1::/common::reiserfs::defaults \
> >> ? ? ? ? ? ? ? Filesystem::/dev/drbd2::/space::reiserfs::defaults,noatime \
> >> ? ? ? ? ? ? ? killnfsd \
> >> ? ? ? ? ? ? ? nfs-common \
> >> ? ? ? ? ? ? ? nfs-kernel-server \
> >> ? ? ? ? ? ? ? Delay::3::0 \
> >> ? ? ? ? ? ? ? IPaddr::xx.xx.xx.xx/23/bond0
> >>
> >> instead, but this doesn't seem to help, either.
> >
> > Do you have the same major/minor numbers exposed to the clients?
> > I think that you can set them in /etc/exports. Otherwise, it
> > seems like nfs-kernel-server in Debian doesn't really stop all
> > nfsd processes (apparently the init.d/nfs-kernel-server has to be
> > fixed, search internet for the relevant links).
> >
> > Thanks,
> >
> > Dejan
> >
> >> Any helpful comments would be very welcome.
> >>
> >>
> >> Regards
> >>
> >> Harri
_______________________________________________
Linux-HA mailing list
Linux-HA [at] lists
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
|
