[ClusterLabs] Antw: Re: Antw: sbd: Cannot open watchdog device: /dev/watchdog

[Klaus Wenninger]

On 01/04/2017 02:23 PM, Muhammad Sharfuddin wrote:
> On 01/04/2017 06:05 PM, Ulrich Windl wrote:
>>>>> Muhammad Sharfuddin <M.Sharfuddin at nds.com.pk> schrieb am
>>>>> 04.01.2017 um 11:58 in
>> Nachricht <9ff82caa-d16e-13f4-e514-d356224f82e7 at nds.com.pk>:
>>> On 01/04/2017 12:09 PM, Ulrich Windl wrote:
>>>>>>> Muhammad Sharfuddin <M.Sharfuddin at nds.com.pk> schrieb am
>>>>>>> 03.01.2017 um 17:11
>>> in
>>>> Nachricht <cdaf4125-f2ce-7793-b4a0-44b2fcaf2ccf at nds.com.pk>:
>>>>> Hello,
>>>>>
>>>>> pacemaker does not start on this machine(Fujitsu PRIMERGY RX2540 M1)
>>>>> with following error in  the logs:
>>>>>
>>>>> sbd: [13236]: ERROR: Cannot open watchdog device: /dev/watchdog:
>>>>> No such
>>>>> file or directory
>>>>>
>>>>> System Info:
>>>>>
>>>>> sbd-1.2.1-8.7.x86_64  corosync-2.3.3-7.12.x86_64
>>>>> pacemaker-1.1.12-7.1.x86_64
>>>>>
>>>>> lsmod | egrep "(wd|dog)"
>>>>> iTCO_wdt               13480  0
>>>>> iTCO_vendor_support    13718  1 iTCO_wdt
>>>> What were the messages when that modules were loaded?
>>>>
>>>> [...]
>>> kernel: [44884.038881] iTCO_vendor_support: vendor-support=0
>>> kernel: [44884.042038] iTCO_wdt: Intel TCO WatchDog Timer Driver v1.11
>>> kernel: [44884.042101] iTCO_wdt: unable to reset NO_REBOOT flag, device
>>> disabled by hardware/BIOS
>>>
>>> what should I check for in firmware ? I have never seen anything in
>>> firmware/bios that relates to watchdog.
>> I don't know for your hardware, but for a HP DL380 iLO is handling
>> the watchdog thing (I think). So maybe look in the corresponding
>> place for your hardware or Google.
> Thank you.
> The unsolved mystery.. on the other 100% identical node(cluster
> member) /dev/watchdog is also unavailable and there I also got the
> same messages when loading iTCO_wdt module but pacemaker starts
> smoothly(even without loading the softdog module).. strange isn't ?

It is - definitely...
You can run sbd without watchdog at all. Maybe this mode is somehow
triggered on your other nodes.
Opening of the watchdog device definitely is always done via a
device-file and not somehow directly
by knowing major/minor of a watchdog device.
Maybe check again how SBD_WATCHDOG_DEV is set (btw. setting it to
/dev/null disables watchdog-use).
On top there is the '-w' parameter to the daemon that can be used to
overrule the '/dev/watchdog'
default and - I think in this order - what comes via environment.


>>>> Ulrich
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> Users mailing list: Users at clusterlabs.org
>>>> http://lists.clusterlabs.org/mailman/listinfo/users
>>>>
>>>> Project Home: http://www.clusterlabs.org
>>>> Getting started:
>>>> http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>>>> Bugs: http://bugs.clusterlabs.org
>>>>
>>>
>>> _______________________________________________
>>> Users mailing list: Users at clusterlabs.org
>>> http://lists.clusterlabs.org/mailman/listinfo/users
>>>
>>> Project Home: http://www.clusterlabs.org
>>> Getting started:
>>> http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>>> Bugs: http://bugs.clusterlabs.org
>>
>>
>>
>>
>> _______________________________________________
>> Users mailing list: Users at clusterlabs.org
>> http://lists.clusterlabs.org/mailman/listinfo/users
>>
>> Project Home: http://www.clusterlabs.org
>> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>> Bugs: http://bugs.clusterlabs.org
>>
>
>
> _______________________________________________
> Users mailing list: Users at clusterlabs.org
> http://lists.clusterlabs.org/mailman/listinfo/users
>
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org