[openib-general] nightly osm_sim report 2006-12-23:normal completion

[Eitan Zahavi]

Hal Rosenstock wrote:
> On Sat, 2006-12-23 at 04:30, Eitan Zahavi wrote:
>   
>> Analysis of the 3 failures:
>> 1. TEST BUG: osmStress - somehow the simulation caused the local port to 
>> be turned down (must be a bug in the random error injection which
>>    should avoid that port). So the simulation ends with OpenSM still 
>> trying to connect to the network.
>> 2. Multicast - The osm.mcfdbs is empty. Apparently no Joins where 
>> received by the SM. This will require further debug.
>>     
I have found a design bug in the queue of MADs waiting for dispatching.
I used an STL map which means that if two MADs where scheduled for the 
exact same time the last time was purging the previous one.
The fix is simple - use multimap instead. It is under testing now.
>> 3. PKey: the test fails on obtaining ALL path records for the 128 node 
>> case. OpenSM complain about timeout during the RMPP transaction.
>>     
>
> Why did this happen now ?
>   
Wish I knew. As I said it is under investigation. Might just be a big 
enough transaction to overflow the 2sec timeout.
>   
>>    I will add some more time to the transaction timeout for the simulation.
>>
>> EZ
>>  
>> Eitan Zahavi wrote:
>>     
>>> OSM Simulation Regression Summary
>>> OpenSM rev = Fri_Dec_22_09:28:50_2006 3ceb7c MOD_FILES=1
>>> ibutils rev = Mon_Dec_18_16:00:49_2006 11d857 
>>> Total=396 Pass=393 Fail=3
>>>
>>> Pass:
>>> 54 Stability IS1-16.topo
>>> 54 Pkey IS1-16.topo
>>> 54 Multicast IS1-16.topo
>>> 54 LidMgr IS1-16.topo
>>> 53 OsmStress IS1-16.topo
>>> 18 Stability IS3-loop.topo
>>> 18 Stability IS3-128.topo
>>> 18 OsmStress IS3-128.topo
>>> 18 Multicast IS3-loop.topo
>>> 18 LidMgr IS3-128.topo
>>> 17 Pkey IS3-128.topo
>>> 17 Multicast IS3-128.topo
>>>
>>> Failures:
>>> 1 Pkey IS3-128.topo
>>> 1 OsmStress IS1-16.topo
>>> 1 Multicast IS3-128.topo
>>>
>>> _______________________________________________
>>> openib-general mailing list
>>> openib-general at openib.org
>>> http://openib.org/mailman/listinfo/openib-general
>>>
>>> To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
>>>   
>>>       
>
>
> _______________________________________________
> openib-general mailing list
> openib-general at openib.org
> http://openib.org/mailman/listinfo/openib-general
>
> To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
>