Re: [Scst-devel] SCST backend device activation problems: scst_translate_lun:FLAG SUSPENDED set, skipping

[Pascal B. <pas...@fr...>]

Right, kernel is being recompiled at the moment. I'll redo a try then.
Thanks for pointing that out, I didn't think the issue mentioned in the
README would produce something like this.

Regards,

Pascal.

-----Message d'origine-----
De : Vladislav Bolkhovitin [mailto:vs...@vl...] 
Envoyé : vendredi 20 juillet 2012 03:04
À : Pascal BERTON
Cc : scs...@li...
Objet : Re: [Scst-devel] SCST backend device activation problems:
scst_translate_lun:FLAG SUSPENDED set, skipping

You log is corrupted. You need to do actions to avoid it described in the
README.

Vlad

Pascal BERTON, on 07/18/2012 01:55 PM wrote:
> Vlad,
>
> I kept digging my issue today, and I had the occasion to have a look 
> at destination node's /var/log/messages file. In fact, for clarity 
> purposes during eventual debug phases on my cluster nodes, I have 
> setup an rsyslog config that redirects various cluster logs from 
> /var/log/messages into dedicated files : drbd.log, scst.log, ra.log, 
> etc... The files I sent yesterday were extracted from scst.log for
instance.
> In normal times, the only entries I used to see in /var/log/messages 
> were minor ntpd ones... But today, when I looked into it for any 
> reason, I discovered very weird things that obviously look directly 
> linked to the symptom we're facing here. It means that what I sent you 
> last night is incomplete, and probably hardly usable to you as is.
> I've extracted the logs corresponding to the time frame of what was 
> related in my 2 yesterday log files, please have a look into it and 
> eventually let me know what you think about it, I've never seen this 
> before. I swear it's not the result of a random character generator, nor
an extract from Matrix.
> Hardware issue in the end ? If not, that would be a very ugly soft one!!!
>
> The best thing to me would be to temporarily redirect the SCST logs 
> back again to /var/log/messages and retry a failing failover so that I 
> can provide you really complete sequence of events, unfortunately the 
> cluster is in production at the moment and my only test volume is 
> currently used to host files that will allow me to empty other volumes 
> and transition them from block dev files to regular xfs files. I may 
> be able to do that tomorrow.
>
> Thanks for your patience and your help anyhow!
>
> Best regards,
>
> Pascal.