SNMP printer discovery documentation?

[Kurt Pfeifle]

John A. Murdie <john at cs.york.ac.uk> wrote (Thursday 27 July 2006 17:54):
 
> I've tried a final experiment. I've stopped the CUPS scheduler and 
> edited printers.conf to remove another printer which snmp finds ok. 
> Then I try 'Administration' -> 'Add New Printer'. The same thing 
> happens as before; I never see a screen like your first screenshot.
> 
> All this is rather painful to do because of the long delays 
> (sometimes 5, sometimes 40 minute) involved in moving from one 
> screen to another in the CUPS management web interface. 

Something's very wrong with your system then... but currently I can't
think of anything that would help to narrow it down. (I'll go back 
and read your previous postings because I can't remember any more all 
the details you've given).

However, I found the snmp (web-based) network printer autodiscovery
also slow (on my self-compiled "trunk" system), while the 1.2.0
release used to be fast, where....

 ..."fast" == less than 3 seconds to finish with discovering 
    22 printers
 ..."slow" == more than 2 minutes (but always less than 3 minutes)
    to discover the same number of printers in the same subnet

Unlike you, I've never experienced it to need 5 or even 40 minutes.

Did you file an STR and attache all the log files that you can 
generate by following the instructions on

   http://www.cups.org/articles.php?L387

? There surely is a way to see just *where* the backend spends that
much time. See f.e. this line (of a CUPS_LOG_LEVEL=2 run):

  DEBUG: 1.150 Scan complete!

The "1.150" string shows it needed a little bit over 1 second. I've
seen a pattern where for *some* printer models CUPS requires 10 or 
more seconds to probe for a supported backend/URL before continuing 
with the next one (which may take less than half a second). And it
looks like it is always the same models.

I'll collect more data and then submit them (need to be back in our
head office for this, though).

Cheers,
Kurt