[ofa-general] Re: OpenSM error codes in a log file

Yevgeny Kliteynik kliteyn at mellanox.co.il
Mon Nov 26 13:34:36 PST 2007


Hal Rosenstock wrote:
> On Mon, 2007-11-26 at 21:28 +0000, Sasha Khapyorsky wrote:
>   
>> Hi!
>>
>> OpenSM when logging an errors uses error code in a log message. Like
>> this:
>>
>> 	osm_log(p_log, OSM_LOG_ERROR,
>> 		"__updn_bfs_by_node (less) ERR AA01: " ...
>> 					       ^^^^
>> This is the old convention, first two digits (AA in this example) are
>> unique per source file, the next two are sequential number of an error
>> in this file.
>>
>> I wanted to know if anybody uses those error codes. If not, are there
>> any plans for using this?
>>
>> IOW should we spend the time and maintain this?
>>     
>
> FWIW, even if no one comes forward with a log parser, I think they are
> useful and should continue to be maintained.
>   
Having a unique error code is a very useful thing - helps
me find the problems reported by users.
Besides, in the OpenSM simulation testing there is kind
of log parser. It's not exactly parsing the log, but it's
monitoring it for "ERR" messages and printing them to the
test log. Again, unique error code is helpful here.

-- Yevgeny

> -- Hal
>
>   
>> Sasha
>>     
> _______________________________________________
> general mailing list
> general at lists.openfabrics.org
> http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
>
> To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
>
>   



More information about the general mailing list