[evla-sw-discuss] CBE-NODE-20 back online

[Martin Pokorny]

Since we're on the subject of bad CBE nodes, we've kept node 29 out of 
use for some time, as well. The problems with it are different than node 
20, as I recall, but I don't remember the details. It's an obvious 
failure, I believe, in that the normal CBE software clearly fails to run 
when that machine is used. Paul would know for certain, but the problem 
doesn't seem to affect the yuppi software. We should remind ourselves 
what's wrong with that node, and fix it so it's possible to bring it 
fully back into service.

On 02/27/2017 03:38 PM, Paul Demorest wrote:
> It actually died within minutes once we started running processing on
> it.  In case it's helpful for debugging, this mode had 128 MB/s coming
> in via two of the 1gig correlator interfaces (p2p1/p2p2), and 0.5 MB/s
> being written out to lustre.
>
> -Paul
>
> On 2017-02-27 15:20, Karyn Roberts wrote:
>> Will do - this is as we expected, though I must admit I'm surprised it
>> only lasted a day.  I'll send the team out tomorrow to retrieve the
>> system for repair.  Thanks for letting me know!
>>
>> Regards,
>> Karyn
>>
>>
>> On 02/27/2017 03:16 PM, Vivek Dhawan wrote:
>>> Node 20 was, up, Paul tried some pulsar processing on it, and it
>>> promptly died. it now seems to have rebooted itself. But we cannot
>>> depend on it during actual observing, so I vote for a real fix.
>>>
>>> We will avoid using it until then.  Priority is moderately high -
>>> a few days without it is OK.
>>>
>>> On Mon, February 27, 2017 11:20, Karyn Roberts wrote:
>>> | The server CBE-NODE-20 is back online - after speaking with K.Scott
>>> | we're not sure for how long - it has crashed twice before and we
>>> expect
>>> | it to crash again under similar circumstances.  Upon the next crash
>>> well
>>> | will remove the service to AOC for service.
>>> |
>>> | Regards,
>>> |
>>> | Karyn
>>> |
>>> | _______________________________________________
>>> | evla-sw-discuss mailing list
>>> | evla-sw-discuss at listmgr.nrao.edu
>>> | https://listmgr.nrao.edu/mailman/listinfo/evla-sw-discuss
>>> |
>>>
>
> _______________________________________________
> evla-sw-discuss mailing list
> evla-sw-discuss at listmgr.nrao.edu
> https://listmgr.nrao.edu/mailman/listinfo/evla-sw-discuss


-- 
Martin