[DRBD-user] DRBD w/ HA w/ Large file issue

pv vishnubhatt at gmail.com
Mon Jul 10 02:05:34 CEST 2006

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


Thx Todd - you're right - I'd incorrectly created the link for /var/lib/nfs
indeed.
Any idea(s): on how to trigger a force sync - if for some reason the two
disks are known to be not-in-sync? is this something that is not recommended
i.e. what if - just to be sure - I set up a script to periodically sync up
two disks - if this were possible, would this cause any untoward
side-effect(s)?
--


On 7/7/06, pv <vishnubhatt at gmail.com> wrote:
>
>  I'm going to ensure that the /var/lib/nfs to make it point to the
> /data/nfs (which is exported and laid on top of log vols), I'm testing that
> as we speak, Thx for the response.
>
> Had yet another question - say for some reason you (as a user/admin)
> realize that the two disks/devs are not in sync - is there a way to force a
> sync from a user-land command?
> --
>
>
>  On 7/6/06, Todd Denniston <Todd.Denniston at ssa.crane.navy.mil> wrote:
> >
> > pv wrote:
> > > I have two servers - fedora1, fedora2, wired up w/ drbd and heartbeat
> > and I'm
> > > exporting the data (residing on the drbd dev) using NFS which is
> > visible via
> > > the cluster interface (via heartbeat).
> > >
> > > The export is visible via a cluster interface (e.g. /data/export). I
> > mount this
> > > export (via the cluster interface) from yet another m/c (nfs client).
> > And I'm
> > > doing simple pings and copying/touching files - when I turn on/off one
> > of the
> > > nodes (e.g fedora1), the mount point moves from fedora1 to fedora2 and
> > back
> > > respectively very well. I copy smaller files and do the same, I do not
> > seem to
> > > have any problem, but when I try copying a large ISO image file, I see
> > the
> > > following issue:
> > >
> > > 1) between the two servers, fedora1 is the primary and I start copying
> > the
> > > large file. at this time - both servers are up and running and copy
> > begins.
> > > 2) in between the copy, I turn off fedora1 and observe that fedora2
> > has taken
> > > over the cluster interface as well as the nfs export/mount point, but
> > I do not
> > > see the large file that I started before the failure.
> >
> > (2) is unexpected by me.
> > are you indicating that the file is not visible at the nfs client or
> > when
> > logged into the server?
> > On the server is a big problem.
> > At the client is probably some minor misconfiguration of the services on
> > the
> > server.
> >
> > > 3) upon failback, I get a stale NFS handle error - in my haresources
> > file, I
> > > have the following entry:
> >
> > what does `ls -ld /var/lib/nfs` (pn both machines) return?
> > The reason I ask is that the data in /var/lib/nfs needs to be on a disk
> > shared
> > between the machines and needs to be accessible _before_ starting nfs
> > and
> > nfslock services.
> >
> >
> > >
> > > fedora1  IPaddr::172.30.30.200/24/eth0 drbddisk::r1
> > > Filesystem::/dev/drbd1::/data::ext3 restart-nfs restart-rpcidmapd
> > >
> >
> > do you control any other related services with ha? like the nfs service?
> >
> > > --
> > >
> > > Any idea if my config is wrong or is there an issue dealing w/ large
> > files
> > > either in drbd or the ha or NFS itself? Thx in advance.
> >
> > not enough info, but I know I have worked with files bigger than 2GB and
> > had
> > no problems on drbd 0.6.13 on Fedora Core 1 over NFS managed by
> > Heartbeat.
> > --
> > Todd Denniston
> > Crane Division, Naval Surface Warfare Center (NSWC Crane)
> > Harnessing the Power of Technology for the Warfighter
> > __
> > please use the "List-Reply" function of your email client.
> >
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20060709/de60e9dd/attachment.htm>


More information about the drbd-user mailing list