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

Re: Ruc130anl


On 5/30/2013 8:48 AM, Dan Swank - NOAA Affiliate wrote:
> Just a few observations and feature requests.
> 
> I noticed that If I shut down TDS servlet from psi-probe the indexing
> continues running until I shut down the tomcat container.    Is this by
> design?   I suppose forcing these threads down by restarting tomcat
> is harmless.   But perhaps put some code in there to shut them
> down gracefully when thredds gets a shutdown signal.

ok, i will check that

> 
> In the cases of monstrous indexing jobs (such as NDFD)
> might be handy to add a configuration item to spect that
> will allow the job to be chopped into multiple process
> threads for the same feature collection.  Say, it scans
> that 20 million  gbx9's need created... fire off 4 threads
> that will work on 5 million each 0-4 5-9 10-14 15-20 M.

i already have a seperate process to do this, configurable with number of 
processes. however one feature collection will only use one process. onviously 
havent fully thought out the needs of these huge collections

if you want to try it out, let me know, heres some prelim docs:

http://www.unidata.ucar.edu/projects/THREDDS/tech/tds4.3/reference/collections/TDM.html

> 
> Increase log rotation size from 1 MB to 10 MB or something configurable.
> I find the NDFD logs are rotating off in a few hours, even without log4j
> DEBUG turned on.   Not enough time to search them for issues.

ok will do

> 
> Unfortunately I didn't look before shuttin' the server down
> this morning (to install new .war) that the two rucanl's were
> still running along side the ndfd.  I'm going to provide these
> logs.. but they may not be of much insight.
> The previous tars I created were done by *.log
> which was leaving out some of the *.log.X rotations..
> this one includes those.
> 
> http://nomads.ncdc.noaa.gov/data/gfsanl/201001-test/tdsLogs-20130530.tar.bz2
>