--> jsafrane has joined #net-snmp
--> [intra]lanman has joined #net-snmp
--> jsafrane has joined #net-snmp
--> jsafrane has joined #net-snmp
--> varun_c has joined #net-snmp
--> intralanman has joined #net-snmp
--> [acer]lanman has joined #net-snmp
--> alienbrain has joined #net-snmp
[2009/06/19 05:35:23] #net-snmp <alienbrain> What's the performance footprint of a "default" net-snmp installation? What can I do to possibly make it as light as possible?
--> hardaker has joined #net-snmp
[2009/06/19 06:33:55] #net-snmp <alienbrain> Does net-snmp retrieve the data periodically even if nobody is asking for it, or does it retrieve it as it's requested?
[2009/06/19 06:35:00] #net-snmp <hardaker> alienbrain: good question; and the answer depends.
[2009/06/19 06:35:16] #net-snmp <hardaker> it depends on the data in question.
[2009/06/19 06:35:31] #net-snmp <hardaker> some of the data it fetches periodicall, but much of it is retrieved on demand.
[2009/06/19 06:36:29] #net-snmp <alienbrain> hardaker: then in that case, would it be beneficial for performance to disable retrieval of such data all together? (assuming it's possible to begin with)
[2009/06/19 06:37:01] #net-snmp <hardaker> well, the only reason that retrieval is sometimes done every so often is if it needs to be.
[2009/06/19 06:37:06] #net-snmp <hardaker> ie, it's never done "just because"
[2009/06/19 06:37:16] #net-snmp <hardaker> so, no...
[2009/06/19 06:37:38] #net-snmp <alienbrain> hardaker: right, I see. that makes sense
[2009/06/19 06:38:47] #net-snmp <alienbrain>
ACTION realizes hardaker joined after he asked his first question and will ask it again, hopping that won't annoy anyone

[2009/06/19 06:38:55] #net-snmp <alienbrain> What's the performance footprint of a "default" net-snmp installation? What can I do to possibly make it as light as possible? :)
[2009/06/19 06:39:21] #net-snmp <hardaker> strip it down to just what you need it to do.
[2009/06/19 06:39:38] #net-snmp <hardaker> see the --with-out-mib-modules flag.
[2009/06/19 06:41:15] #net-snmp <alienbrain> hardaker: there is really nothing specific I want to do, I'm just trying to understand the general performance of net-snmp and what, if any, are the sacrifices I can make in order to improve it
[2009/06/19 06:41:32] #net-snmp <alienbrain> sacrifices in features, that is
[2009/06/19 06:42:04] #net-snmp <alienbrain> hardaker: I will take a look at the flag
[2009/06/19 06:42:09] #net-snmp <hardaker> http://net-snmp.sourceforge.net/FAQ.html#How_can_I_reduce_the_memory_footprint_
[2009/06/19 06:42:15] #net-snmp <hardaker> Also, take a look at the FAQ in general.
[2009/06/19 06:42:24] #net-snmp <hardaker> it answers a number of your questions
[2009/06/19 06:42:39] #net-snmp <alienbrain> Very nice!
[2009/06/19 06:42:42] #net-snmp <alienbrain>
ACTION reads

[2009/06/19 06:49:20] #net-snmp <alienbrain> hardaker: thanks very much
[2009/06/19 06:49:29] #net-snmp <hardaker> no problem.
[2009/06/19 06:53:16] #net-snmp <alienbrain> The conclusion I understand so far, is that MIBs are loaded but most of them are not doing anything except occupying memory, which on my system seem to be 8Mb of physical memory. That's not bad at all. Even if I start pulling more data and it goes up for as much as 16 Mb, that would be cool.
[2009/06/19 06:53:33] #net-snmp <alienbrain> modules that implement MIBs are loaded, that is
[2009/06/19 06:58:50] #net-snmp <hardaker> modules are optional components that implement MIBs
[2009/06/19 06:59:12] #net-snmp <hardaker> they take memory just to be loaded (code size) and memory when run (especially if they have a cache)
[2009/06/19 06:59:28] #net-snmp <hardaker> and CPU power to process the results (which you won't care about if you never query them for the results anyway)
[2009/06/19 06:59:38] #net-snmp <rstory> alienbrain: you will want to disable the notification log saving traps (or disable it entirely).. that will eat lots of memory over time
[2009/06/19 07:00:11] #net-snmp <rstory> at the very least reduce the default number to save from 1000 to something much smaller
[2009/06/19 07:01:01] #net-snmp <alienbrain> rstory: is that related to traps? I actually don't need traps at all (if I understand correctly, they are used to write to SNMP), and I'd rather disable them all together
[2009/06/19 07:02:55] #net-snmp <alienbrain> rstory: didn't notice you said traps already :)
[2009/06/19 07:03:53] #net-snmp <alienbrain> rstory: I don't have snmptrapd running, is there anything else I should do to disable it?
[2009/06/19 07:04:38] #net-snmp <rstory> traps aren't used to write..
[2009/06/19 07:04:47] #net-snmp <rstory> they are used to send notifications of events
[2009/06/19 07:04:59] #net-snmp <rstory> for example, the agent sends a trap when it starts or stops
[2009/06/19 07:05:21] #net-snmp <rstory> and the agent keeps a log of which traps it has sent.. that's what I was referring to..
[2009/06/19 07:05:27] #net-snmp <rstory> nothing to do with snmptrapd
[2009/06/19 07:08:07] #net-snmp <alienbrain> rstory: ah I see. I will check snmpd.conf for how to adjust log size accordingly
[2009/06/19 07:08:32] #net-snmp <alienbrain> hardaker: that relieving to know!
--> kisielk has joined #net-snmp
--> [intra]lanman has joined #net-snmp
--> fatpelt_laptop has joined #net-snmp
[2009/06/19 09:48:35] #net-snmp <fatpelt_laptop> afternoon all. I'm trying to run snmptrapd with a traphandle default exactly as is written in the man page for snmptrapd.conf and it never gets called. Do i have to enable something else somewhere?
--> hardaker2 has joined #net-snmp
--> T_X has joined #net-snmp
--> frostburn has joined #net-snmp
--> hardaker has joined #net-snmp
[2009/06/19 12:34:12] #net-snmp <fatpelt_laptop>
ACTION tracked his issue down to authentication. he didn't realize perms were different when using a traphandler

--> Hobart has joined #net-snmp
--> T_X_ has joined #net-snmp
--> mike_irc has joined #net-snmp
--> T_X has joined #net-snmp
--> hardaker has joined #net-snmp
--> Hobart has joined #net-snmp
--> Hobart_ has joined #net-snmp
[2009/06/19 14:03:28] #net-snmp <Hobart> anyone know how to get around the spraying of several "Cannot find module (IP-MIB): At line 0 in (none)" errors if you try to pre-load a specific mib?
[2009/06/19 14:11:10] #net-snmp <Hobart> hardaker -> around? :)
[2009/06/19 14:16:28] #net-snmp <hardaker> yep.
[2009/06/19 14:16:30] #net-snmp <hardaker> ish
[2009/06/19 14:17:04] #net-snmp <hardaker> that usually means it can't load your mibs because they're not installed or not in the place they're expected.
[2009/06/19 14:19:54] #net-snmp <Hobart> that's definitely the case, but I'm trying right-off-the-bat to tell it where they are, my first invocation of the netsnmp stuff is an addMibDirs() call to where they all are
[2009/06/19 14:21:56] #net-snmp <Hobart> I'm watching it in process exporer, and I see the process try to look at HKCU(lm)\Software\Net-SNMP, then look for \usr\etc \usr\share \usr etc etc
[2009/06/19 14:22:07] #net-snmp <Hobart> I tried telling perl to set a MIBDIRS env variable first, no luck
[2009/06/19 14:25:07] #net-snmp <hardaker> err... run with full debugging to see where it's looking.
[2009/06/19 14:26:45] #net-snmp <Hobart> heh, it'd be nice if I wasn't using a binary I guess :)
[2009/06/19 14:26:55] #net-snmp <hardaker> no,
[2009/06/19 14:27:05] #net-snmp <hardaker> what program are you actually running?
[2009/06/19 14:27:07] #net-snmp <hardaker> one of ours?
[2009/06/19 14:27:09] #net-snmp <hardaker> ie, snmpget?
[2009/06/19 14:27:16] #net-snmp <hardaker> if so, run with -Dall
[2009/06/19 14:27:19] #net-snmp <Hobart> per http://net-snmp.svn.sourceforge.net/viewvc/net-snmp/trunk/net-snmp/win32/dist/htmlhelp/Configuration_Overview.html I was about to try setting SNMPSHAREPATH , should that affect where it looks for mibs?
[2009/06/19 14:27:23] #net-snmp <hardaker> and be prepared to run away with the data you get.
[2009/06/19 14:27:24] #net-snmp <Hobart> hardaker: what I'm running, is actually
[2009/06/19 14:27:44] #net-snmp <Hobart> a compiled-with-par .exe that's running win32 perl with net snmp's dlls bundled :)
[2009/06/19 14:27:52] #net-snmp <hardaker> ah.
[2009/06/19 14:28:00] #net-snmp <hardaker> then make a "snmp.conf" file
[2009/06/19 14:28:02] #net-snmp <hardaker> and in it put:
[2009/06/19 14:28:13] #net-snmp <hardaker> debugTokens ALL
[2009/06/19 14:28:16] #net-snmp <hardaker> doDebugging 1
[2009/06/19 14:28:30] #net-snmp <hardaker> anyway, I gotta split now... I'll be around later though.
[2009/06/19 14:28:33] #net-snmp <Hobart> will it look in the running directory by default?
[2009/06/19 14:28:36] #net-snmp <Hobart> hardaker: many thanks!
[2009/06/19 14:28:51] #net-snmp <hardaker> no, it needs to be in the right place.
[2009/06/19 14:29:11] #net-snmp <hardaker> but setting the SNMPCONFPATH environment variable to "." would probably work.
--> mike_irc has joined #net-snmp
--> hardaker has joined #net-snmp
--> varun_c has joined #net-snmp
--> mike_irc has joined #net-snmp
--> Hobart has joined #net-snmp
--> T_X has joined #net-snmp
--> varun_c has joined #net-snmp
--> varun_c_ has joined #net-snmp
--> varun_c_ has joined #net-snmp
--> varun_c_ has joined #net-snmp
--> hardaker has joined #net-snmp
--> asir has joined #net-snmp
[2009/06/19 20:37:58] #net-snmp <asir> hi! my MIB compiler truncates all OIDs to unsigned int, is this behavior related to some RFC?
--> magfr has joined #net-snmp
--> magfr has joined #net-snmp
--> varun_c_ has joined #net-snmp
--> af_ has joined #net-snmp
--> alienbrain has joined #net-snmp
--> snmpbot has joined #net-snmp