[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

20030327: 20030327: rtstats issues

Carl,

Yes, but in essence, the ldm.pq is already a data base (circular).
Rtstats is just reading ldm.pq information and creating an LDM
product synopsis for transmission. On this end, I am sending all the
received stats into NetCDF, but SQL or something else could be used.
To just log all products without sending across the net, it would
be better to just interface the queue scanning directly with the 
output database.

Steve Chiswell 

>From: Carl Sinclair <address@hidden>
>Organization: CAPS
>Keywords: 200303271741.h2RHfOB2007493

>Steve,
>
>Has any thought been put into having rtstats write directly to a
>database instead of to a text file?
>
>Carl
>
>
>> Carl,
>> 
>> The origin_v_upstream was added tp the LDM in 5.2 as part of the product
>> structure (eg _v_upstream is appended to the origin hostname). It is not
>> being added by rtstats. Each LDM will replace the _v_upstream string with th
> e
>> hostname of the machine that the product was received from.
>> 
>> Prior to LDM5.2, the only way to determine topology was for sites running
>> mailpqstats from their cron to examine their ldmd.log file for "topo" entrie
> s.
>> Unfortunately, the "topo" line in ldmd.log is only emitted upon the
>> initial connection, making the topology information very static.
>> The addition of the upstream information allows the statistics to determine
>> what the topology is at any moment, including if there are multiple
>> paths to a site. Eventuellay, we would like to improve IDD routing
>> using this information.
>> 
>> Steve Chiswell
>>
>-- 
>_______________________
>Carl Sinclair
>OU/CAPS Research Associate
>(405)325-5938
>