Re: [SSI-devel] [DRBD] drbd-0.7.5-ssi_pre15 for SSI-1.1.1-FC2

[Roger T. <pe...@ho...>]

Sorry it's been a few weeks ago so I don't recall exactly.  It probably
was trivial because I didn't experience a problem setting up.  It is
highly likely I simply manually --do-what-I-say in bash when the linuxrc
failed to set primary the first time rebooting.  Then mounted root.  Try
that.


>> I resized an existing ext3 root partition to make room for the drbd
>> journal.  Then do drbdadm attach and the drbdsetup --do-what-I-say.
>
> Was the root partition mounted when you did this? Because when I call
> 'drbdadm attach r0', I get this:
>
> [root@lx19944 root]# drbdadm attach r0
> Lower device (/dev/hda6) is already mounted
> Command '/sbin/drbdsetup /dev/drbd/0 disk /dev/hda6 /dev/hda7 0
> --on-io-error=detach' terminated with exit code 20
> drbdadm aborting
> [root@lx19944 root]#
>
> Regards,
>
> En Chiang
>
>>   I
>> believe you don't have to drbdadm up or connect to make it set to
>> primary.
>>  Of course do this before you reboot, so you don't have to modify your
>> linuxrc to force primary.  And of course you should reboot before you
>> sync.
>>
>>
>> > Hi Roger,
>> > I finally got down to testing your patch for root failover, and I have
>> a
>> question about the config you've been testing.
>> >  - Did you set up your failover root on an existing mounted / fs or
>> did
>> > you build another root which you mirrored over DRBD?
>> > The reason I ask this is that for drbd 0.7.5, the first time you
>> 'drbdadm up' a resource, you cannot set it to primary since it is
>> 'Inconsistent' and you have to use the '--do-what-I-say' option. Now,
>> this is not possible on a mounted partition, since 'drbdadm up' itself
>> will fail.
>> > This becomes a problem when I attempt to layer drbd over an existing
>> root. I have to add the --do-what-I-say option to the linuxrc to mark
>> the drbd device consistent for the first potential root node to boot.
>> This is okay for the first boot of the cluster... but for a subsequent
>> boot, if the node booting up first was actually inconsistent (an
>> incomplete sync or something like that), then forcing it to mark itself
>> consistent could cause a serious problem in the cluster. It should
>> actually wait for the mirror to come up first!
>> > Thanks,
>> > En Chiang
>> > On Sun, 2004-11-14 at 10:13, Roger Tsang wrote:
>> >> Patch this against the original drbd-0.7.5 source to use the latest
>> DRBD
>> >> with SSI-1.1.1.
>> >> Lotsa testing but only on Fedora Core 2.  Lotsa fixes.
>> >> Enjoy.
>> >> -Roger
>> > -------------------------------------------------------
>> > SF email is sponsored by - The IT Product Guide
>> > Read honest & candid reviews on hundreds of IT Products from real
>> users.
>> Discover which products truly live up to the hype. Start reading now.
>> http://productguide.itmanagersjournal.com/
>> > _______________________________________________
>> > ssic-linux-devel mailing list
>> > ssi...@li...
>> > https://lists.sourceforge.net/lists/listinfo/ssic-linux-devel
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> -------------------------------------------------------
>> SF email is sponsored by - The IT Product Guide
>> Read honest & candid reviews on hundreds of IT Products from real users.
>> Discover which products truly live up to the hype. Start reading now.
>> http://productguide.itmanagersjournal.com/
>> _______________________________________________
>> ssic-linux-devel mailing list
>> ssi...@li...
>> https://lists.sourceforge.net/lists/listinfo/ssic-linux-devel
>>
>
>
>
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from real users.
> Discover which products truly live up to the hype. Start reading now.
> http://productguide.itmanagersjournal.com/
> _______________________________________________
> ssic-linux-devel mailing list
> ssi...@li...
> https://lists.sourceforge.net/lists/listinfo/ssic-linux-devel
>