Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Wed, Jun 13, 2007 at 02:30:38PM -0400, Craig Hoffman wrote:
>
>
> Lars Ellenberg wrote:
> >On Tue, Jun 12, 2007 at 01:51:04PM -0400, Craig Hoffman wrote:
> >
> >>I'm having a problem getting my secondary node to see that data has
> >>changed on the primary and that it needs to sync up.
> >>The problem seems to be happening to me on both 0.7 and 0.8. I'm sure
> >>that it's just something that I'm doing, but I'd like a little
> >>
> >pointer.
> >
> >>Here's what I've done to replicate the problem...
> >>
> >>Bring up both nodes. Force one into being a primary. The secondary
> >>node notices this and they start syncing up. This happened when I
> >>
> >first
> >
> >>set DRBD up. If I leave things be, the two seem to remain in sync
> >>
> >just
> >
> >>fine. Catting /proc/drbd shows the numbers changing. Good. But, if
> >>
> >I
> >
> >>do a /etc/init.d/drbd stop and then a start on the secondary, things
> >>never sync again *unless* I force the secondary as inconsistent.
> >>
> >>My backend devices are LVM. So, in effect, I'm running drbd on top of
> >>LVM. My primary server names are xen04 and my secondary is xen05.
> >>
> >>Here is a snippet from my drbd.conf:
> >>
> >>resource logserver {
> >>protocol C;
> >>startup {
> >> degr-wfc-timeout 120;
> >>}
> >>disk {
> >> on-io-error detach;
> >>}
> >>net {}
> >>syncer {
> >> rate 10M;
> >>}
> >>on xen05 {
> >> device /dev/drbd0;
> >> disk /dev/mapper/domu-logserver;
> >> address 192.168.0.60:7788;
> >> meta-disk /dev/domu/drbdmeta[1];
> >>}
> >>on xen04 {
> >> device /dev/drbd0;
> >> disk /dev/mapper/domu-logserver;
> >> address 192.168.0.72:7788;
> >> meta-disk /dev/domu/drbdmeta[1];
> >>}
> >>
> >
> >you are sure you want to have a drbd between two domU ?
> >
> >you are sure that the "disk = []" in you xen configs is ok,
> >no copy'n'paste error? (sorry, I have to ask, that _does_ happen)
> >
> I believe that I'm sure. ;)
> The domU on xen05 is not operational. It is meant to be a cold failover.
>
> The disk line in my xen config file looks like this:
> disk = [ 'phy:domu/logserver,hda1,w','phy:domu/logserver_swap,hda2,w' ]
soo...
why does that not read phy:drbd0, then?
> To further elaborate, my xen config works just fine. And when I first
> setup drbd, the LV's were mirrored as I expected them to - so I assume
> that my configs are fine.
> >
> >
> >>My /proc/drbd earlier in the day:
> >>
> >>xen05:/usr/src/drbd-8.0.3# cat /proc/drbd
> >>version: 8.0.3 (api:86/proto:86)
> >>SVN Revision: 2881 build by root at xen05, 2007-06-11 11:57:14
> >>0: cs:Connected st:Secondary/Primary ds:UpToDate/UpToDate C r---
> >> ns:0 nr:20480000 dw:20480000 dr:0 al:0 bm:1250 lo:0 pe:0 ua:0 ap:0
> >> resync: used:0/31 hits:1278750 misses:1250 starving:0 dirty:0
> >>changed:1250
> >> act_log: used:0/127 hits:0 misses:0 starving:0 dirty:0
> >>
> >changed:0
> >
> >>
> >>And then after I've done a
> >>/etc/init.d/drbd stop
> >>/etc/init.d/drbd start
> >>on the secondary machine.
> >>
> >>xen05:/usr/src/drbd-8.0.3# cat /proc/drbd
> >>version: 8.0.3 (api:86/proto:86)
> >>SVN Revision: 2881 build by root at xen05, 2007-06-11 11:57:14
> >>0: cs:Connected st:Secondary/Primary ds:UpToDate/UpToDate C r---
> >> ns:0 nr:0 dw:0 dr:0 al:0 bm:0 lo:0 pe:0 ua:0 ap:0
> >> resync: used:0/31 hits:0 misses:0 starving:0 dirty:0 changed:0
> >> act_log: used:0/127 hits:0 misses:0 starving:0 dirty:0
> >>
> >changed:0
> >
> >what makes you think there is something wrong with this output?
> >
> Well, prior to stopping the DRBD service on xen05, I saw those numbers
> change. Now all of them show "0". "Hit's" for example - prior to
> stopping the service, there were 1278750 hits. After I stopped the
> service and restarted it, my hits stayed at 0.
forget about the "hits" and that stuff...
the interesting counters are
ns:0 nr:0 dw:0 dr:0 al:0 bm:0
> Also, if I stop the
> service, mount that LV and look at the data, nothing new is in there.
> If I reconnect my DRBD devices, and let them sit for a month, even
> though it says that they're UpToDate, they're not. If Xen04 (my
> primary) were to die, I would not be able to bring xen05 up with current
> data. The data would only be current as of the last time I manually
> said that the secondary(xen05) was inconsistent. I could write a script
> that would do this once a day, but that's not much better than a typical
> backup.
I really don't know what is wrong with your config, but that sounds
really broken, and I doubt it is drbd.
maybe you bypass drbd with your writes?
your disk line in your xen config point in that direction, too.
I have no idea why xen would allow that, but aparently it does....
--
: Lars Ellenberg Tel +43-1-8178292-0 :
: LINBIT Information Technologies GmbH Fax +43-1-8178292-82 :
: Vivenotgasse 48, A-1120 Vienna/Europe http://www.linbit.com :
__
please use the "List-Reply" function of your email client.