Best way for NFS failover - SUN

This is a discussion on Best way for NFS failover - SUN ; I am in the process of setting up HA failover in our environment. So far, the only clustered components are two Cisco content switches. These will direct traffic to servers (Solaris 8/9/10) on their LAN (round robin with automatic failover), ...

+ Reply to Thread
Results 1 to 6 of 6

Thread: Best way for NFS failover

  1. Best way for NFS failover

    I am in the process of setting up HA failover in our environment.
    So far, the only clustered components are two Cisco content switches.
    These will direct traffic to servers (Solaris 8/9/10) on their LAN
    (round robin with automatic failover), and the servers in turn have NFS
    mounted a NAS storage box.

    We have two NAS units (Adaptec Snapservers), and the primary NAS unit
    automatically syncs its data to the standby NAS unit so we can have an
    up to the minute mirror of the data on the primary box.

    What I need to do, is have the servers automatically NFS mount the
    second NAS unit if the first one fails. I'm not sure if this is a
    feature in NFS or if a simple script can solve this; something that
    checks for a file on the primary NAS unit, if it hangs in the process or
    is unable to find it, it umount -f the primary NAS unit and mounts the
    seconday NAS unit.

    I really don't know what the best solution is, a simple script is the
    easiest way, but is there any other, better and perhaps more robust and
    elegant ways?

    --
    Stig Bull
    | remove .no.spam from my email address to reply by mail |
    No animals were hurt or killed in the process of creating this
    electronic message. To reduce download time, this message is made of
    100% recycled bytes.

  2. Re: Best way for NFS failover

    Stig Bull wrote:
    > What I need to do, is have the servers automatically NFS mount the
    > second NAS unit if the first one fails. I'm not sure if this is a
    > feature in NFS or if a simple script can solve this; something that
    > checks for a file on the primary NAS unit, if it hangs in the process or
    > is unable to find it, it umount -f the primary NAS unit and mounts the
    > seconday NAS unit.
    >
    > I really don't know what the best solution is, a simple script is the
    > easiest way, but is there any other, better and perhaps more robust and
    > elegant ways?



    Hi, when your NAS unit switch for a failover, the ip address for mount
    the filesystem is the same of the NAS unit that have crashed?
    If yes, i think you need only to mount your fiesystem with soft option,
    so df -k and other comand don't will stop during filesystem unavailable.

    The only unsure think it's the mac address of network adapter of NAS
    units, if it's the same sure you'll don't have problems for a failover,
    but if the network adapter have a different mac address, maybe the
    server can need more time to give back filesystem.

    Hope to help you...


    Fabrizio

  3. Re: Best way for NFS failover

    In article , famany@famany.it
    says...

    > Hi, when your NAS unit switch for a failover, the ip address for mount
    > the filesystem is the same of the NAS unit that have crashed?
    > If yes, i think you need only to mount your fiesystem with soft option,
    > so df -k and other comand don't will stop during filesystem unavailable.
    >
    > The only unsure think it's the mac address of network adapter of NAS
    > units, if it's the same sure you'll don't have problems for a failover,
    > but if the network adapter have a different mac address, maybe the
    > server can need more time to give back filesystem.


    Hi, thanks for your reply.

    I forgot to add, the NAS units have different IP addresses, they are not
    clustered in any other ways other than the primary unit mirrors its data
    to the secondary unit.

    However I've done some Googling, and the best way for our technical
    solution, is a script which umounts/mounts the NAS units.

    --
    Stig Bull
    | remove .no.spam from my email address to reply by mail |
    No animals were hurt or killed in the process of creating this
    electronic message. To reduce download time, this message is made of
    100% recycled bytes.

  4. Re: Best way for NFS failover

    Stig Bull wrote:
    > I forgot to add, the NAS units have different IP addresses, they are not
    > clustered in any other ways other than the primary unit mirrors its data
    > to the secondary unit.
    >
    > However I've done some Googling, and the best way for our technical
    > solution, is a script which umounts/mounts the NAS units.


    Yes i think that if your NAS unit haven't a failover ip, the only
    solution is a simple script. If do you have problem for write it, don't
    warry ask to us

    Bye

    Fabrizio I.

  5. Re: Best way for NFS failover

    Fabrizio Imbroglia wrote:
    > Stig Bull wrote:
    >> I forgot to add, the NAS units have different IP addresses, they are not
    >> clustered in any other ways other than the primary unit mirrors its data
    >> to the secondary unit.
    >>
    >> However I've done some Googling, and the best way for our technical
    >> solution, is a script which umounts/mounts the NAS units.

    >
    > Yes i think that if your NAS unit haven't a failover ip, the only
    > solution is a simple script. If do you have problem for write it, don't
    > warry ask to us


    No, Solaris allows "replicated" NFS mounts, see man mount_nfs.
    If these are used with automount, you can even specify priorities for
    the servers, see man automount. But all this is restricted to read-only
    mounting.

    --

    Hans Werner Strube strube(@)physik3(.)gwdg(.)de
    Drittes Physikalisches Institut, Univ. Goettingen
    Friedrich-Hund-Platz 1, 37077 Goettingen, Germany

  6. Re: Best way for NFS failover

    In article , me@privacy.net says...

    > No, Solaris allows "replicated" NFS mounts, see man mount_nfs.
    > If these are used with automount, you can even specify priorities for
    > the servers, see man automount. But all this is restricted to read-only
    > mounting.


    Thanks for your input, all, I'll stick with a simple script; I read the
    mount_nfs man pages and like you say, that's only if the mount is ro.

    --
    Stig Bull
    | remove .no.spam from my email address to reply by mail |
    No animals were hurt or killed in the process of creating this
    electronic message. To reduce download time, this message is made of
    100% recycled bytes.

+ Reply to Thread