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

[LDM #DBK-582040]: 6.10.1 how to diagnose pqact memory leak?



Daryl,

> Thanks for the response, if you could induldge me with one more round of
> questions, please see inline....
> 
> So I am now confused about the need to split pqact processes up for high
> volume feeds, like NEXRAD2 .  I thought the justification was that if you
> have 80 some RADARs producing data and thus writing to files, pqact would
> have to be constantly cycling file descriptors so to only have 32 open at
> once? I thought the split was done to limit the number of RADARs per
> pqact to something less than 32, so that FDs are kept open and recycled
> more slowly...

I don't recall ever hearing that rationale.

> Or is the split done so that there could be multiple write processes
> ongoing at one time?

Reasons for splitting in decreasing order of importance (in my opinion):

    1. To make things easier to understand and maintain (all radar processing 
in one pqact(1) file, etc.);
    2. To take advantage of multiple CPU-s;
    3. To reduce file-descriptor reuse.

> Thanks, I have been attempting to understand the "twisted
> logic" of close_lru() and your statement above helps in that! :)

Glad I could help.

> thanks again,
> daryl
> 
> --
> /**
> * Daryl Herzmann
> * Assistant Scientist -- Iowa Environmental Mesonet
> * http://mesonet.agron.iastate.edu

Regards,
Steve Emmerson

Ticket Details
===================
Ticket ID: DBK-582040
Department: Support LDM
Priority: Normal
Status: Closed