[DRBD-user] dmesg: drbd_bm_set_bits_in_irq ?
Matteo Tescione
matteo at rmnet.it
Tue Dec 11 03:55:50 CET 2007
Hi Vincent,
Il 10-12-2007 18:55, "Fortier,Vincent [Montreal]"
<Vincent.Fortier1 at EC.GC.CA> ha scritto:
>> -----Message d'origine-----
>> De : drbd-user-bounces at linbit.com
>> [mailto:drbd-user-bounces at linbit.com] De la part de Florian Haas
>>
>> Vincent,
>>
>> Problem reproduced. Steps to reproduce:
>>
>> 1. drbdadm down <resource>
>> 2. remove underlying device (e.g. rename logical volume) 3.
>> drbdadm up <resource>
>>
>> after "drbdadm up", DRBD issues this error message:
>>
>> Can not open device '<local device>': No such file or
>> directory Command 'drbdsetup /dev/drbd<num> disk <local
>> device> <local device> internal --set-defaults
>> --create-device --on-io-error=detach' terminated with exit code 20
>>
>> This is likely to confuse users. This doesn't mean that the
>> "up" command failed altogether. "up" is really the
>> combination of "detach"
>> and "connect"; "detach" fails here while "connect" proceeds
>> as normal.
>>
>> The device is then Diskless. But, the in-memory
>> representation of the sync bitmap was never initialized.
>>
>> 4. drbdadm primary <resource>
>> 5. issue any write I/O to DRBD device.
>>
>> When write I/O is issued after the device has been made
>> Primary, DRBD attempts to update an in-memory sync bitmap
>> that isn't there. Hence your error messages in the kernel log.
>>
>> This is a bug. A fix is being worked on. It is not fixed in 8.0.8.
>
> Glad to know I've shake-up things to make this bug emerge and get
> resolved...
>
>> Still, you would never have run into this issue if your
>> underlying disks hadn't been totally broken (as they were,
>> apparently).
>
> Apparently they where... although reasons unknown...
>
I occasionally had this kind of behaviour when using lv's over drbd devices,
and i found it was related of the filters inside lvm.conf, telling lvm stuff
to lookup in drbd devices, breaking the underlying meta-data.
See if this helps,
Cheers,
--
#Matteo Tescione
#RMnet srl
>> From your last message, it appears that you have fixed your
>> setup, and also ran a full resync to be on the safe side.
>> Which is a good thing.
>
> Yeah full resynch is on go... drbd0 is done and drbd1 is now at 50%.
>
> thnx for the info.
>
> - vin
> _______________________________________________
> drbd-user mailing list
> drbd-user at lists.linbit.com
> http://lists.linbit.com/mailman/listinfo/drbd-user
>
More information about the drbd-user
mailing list