Re: [pysnmp-users] cmdgen and SNMPv3

[Vincent B. <be...@lu...>]

 ❦ 18 novembre 2015 21:44 +0100, Ilya Etingof <il...@gl...> :

> Oh, that might be a problem then. I’ve not noticed that advice.
>
> The good thing is that the parsing phase only happens once, then the
> result is put into $HOME/.pysnmp/mibs/ and only loaded from there.
> So you can potentially ‘bootstrap’ all MIBs from a single thread to
> compile them from ASN.1 source prior to use (there’s also the mibdump
> command-line tool).

Currently, I am not using any MIB myself. I am only using OIDs. The only
MIB that I am using are those implicitly needed by the SNMP engine. I am
quite unsure of what happens exactly.

$HOME/.pysnmp doesn't exist on my system, so I suppose that no MIB were
cached. I may be mistaken with my view on how all this work. I was under
impression that instantiating a new SnmpEngine would read some MIB (and
that was the part that was slow).

> Or have a dedicated thread that only compiles newly required MIBs,
> other threads never compile anything but only load them up.
>
> I do not see much better solutions since we seem to hit ply’s limitation.
>
> What do you think?

Yes, this is a very reasonable limitation (loading the MIB before doing
thread stuff).
-- 
The first thing we do, let's kill all the lawyers.
		-- Wm. Shakespeare, "Henry VI", Part IV