Re: flexfax: Problems submitting very large Number of jobs sequential ly

[Nico Kadel-Garcia <raoul@cirl.meei.harvard.edu>]

-----BEGIN PGP SIGNED MESSAGE-----

On Thu, 23 Jul 1998, Christoph Stotz wrote:

> Hello !
> 
> We are using hylafax to send information to our customers. Normally
> about 500 faxes/day.

Cool. What operating system, Linux? And what version of HylaFAX?

> >From time to time we have to send more than 5000/day. Before sending we
> generate all the .tif´s and
> then call a batch-programm that submit´s all the faxes sequentially.
> After some time
> submitting becomes slower and slower and CPU-Usage by faxq gets higher:
> 
> ps -aux | grep faxq
> 
> uucp      5139 24.0  2.1  2552  1348  ?  R    Jul 21 631:16
> /usr/sbin/faxq

Umm. Could we see the batch script?

> We are using 200MHz Pentium Pro with 64MB of RAM and 12GB of Harddisk,
> so this
> should not be the problem. After 1500 jobs in the queue, the time
> between two submitted jobs
> is about 7-10 seconds.

*1500 faxes in the queue*? Yowzah! How big is your spool partition, and
could it be getting over-stuffed and slowing down?

> PS: I´m suspecting that it could by a "low memory" problem, but I do not
> understand whay hylafax
> should use 50 MB of RAM for 1500 jobs ???

Umm. I assume you have a good chunk of swap space enabled? And are
you running hfaxd from inetd or as a stand-alone daemon?

Also, each job is (I think) opening a connection to hfaxd: you could
be vastly over-loading your available TCP connections....


			Nico Garcia
			Senior Engineer, CIRL 
			Mass. Eye and Ear Infirmary
			raoul@cirl.meei.harvard.edu

-----BEGIN PGP SIGNATURE-----
Version: 2.6.2

iQCVAwUBNbdaIT/+ItycgIJRAQExlgP/bpsIpwv6s3xfRwDXXfLj11aajIcJt3Qn
mnIBrD1YDz8G6eM5sNJ8aGAMiX98UWWzmssAYQbLXG9xUfqVU5tSs8iS2u79cdvC
pGuZlEoBb0TbIRhAiqXo9zFAcbWU9XMvxVFQmZufYafbdbE5GO1gkfqVvf+UEm3a
iDIG7oA4iAw=
=WtzM
-----END PGP SIGNATURE-----