--> jbpn has joined #net-snmp
--> MezoWork has joined #net-snmp
[2002/10/22 04:58:28] #net-snmp <MezoWork> hi all
[2002/10/22 07:19:00] #net-snmp <MezoWork> rstory: have U read this mail : need explanation about how run table_iterator API ?
[2002/10/22 07:19:31] #net-snmp <rstory> yes, but I can't help. Wes wrote the table_iterator.
[2002/10/22 07:19:38] #net-snmp <MezoWork> ok ...
[2002/10/22 07:19:54] #net-snmp <MezoWork> doesn't understand why snmpd browse all my table ...
[2002/10/22 07:20:17] #net-snmp <MezoWork> it doesn't stop at the end of the table which was requested ...
[2002/10/22 07:21:38] #net-snmp <rstory> if you are using snmp-get, it is searching for the next object...
[2002/10/22 07:24:23] #net-snmp <MezoWork> I use snmptable
[2002/10/22 07:24:28] #net-snmp <MezoWork> to make my request
[2002/10/22 07:25:52] #net-snmp <rstory> and snmptable uses get-next requests...
[2002/10/22 07:27:40] #net-snmp <MezoWork> in code ...
[2002/10/22 07:27:52] #net-snmp <MezoWork> get_next is like get
--> wes has joined #net-snmp
[2002/10/22 08:20:49] #net-snmp <MezoWork> hi wes
[2002/10/22 08:20:50] #net-snmp <MezoWork> wes have read this mail : need explanation about how run table_iterator API ... ?
[2002/10/22 08:23:06] #net-snmp <wes> what's the question?
[2002/10/22 08:24:38] #net-snmp <rstory> wes: we're getting quite a few questions about memory usage w/table_iterator these days...
[2002/10/22 08:24:54] #net-snmp <wes> Yeah, I need to take a whack at it again.
[2002/10/22 08:24:57] #net-snmp <MezoWork> well ...
[2002/10/22 08:25:07] #net-snmp <MezoWork> when I make a request on a table
[2002/10/22 08:25:07] #net-snmp <wes> Now that I have a decent test candidate (pluto) I'll see what I can do today.
[2002/10/22 08:25:12] #net-snmp <MezoWork> with snmptable ...
[2002/10/22 08:25:33] #net-snmp <MezoWork> snmpd browse all table ... and the memory grow up ...
[2002/10/22 08:25:38] #net-snmp <MezoWork> for example
[2002/10/22 08:25:44] #net-snmp <MezoWork> if I request ifTable ...
[2002/10/22 08:25:49] #net-snmp <wes> rstory: I actually never finished writing it either (as noted by all the ifdefs)
[2002/10/22 08:26:12] #net-snmp <MezoWork> snmpd browse ifTable and all the following table ....
[2002/10/22 08:26:27] #net-snmp <MezoWork> why browse the following table ?
[2002/10/22 08:26:29] #net-snmp <rstory> hey, I'll be the last one to throw stones from within my own glass house!
[2002/10/22 08:27:30] #net-snmp <MezoWork> wes ... ?
[2002/10/22 08:27:55] #net-snmp <rstory> MezoWork: be patient...
[2002/10/22 08:27:56] #net-snmp <wes> it shouldn't browse the following table.
[2002/10/22 08:28:08] #net-snmp <wes> (I can't type that fast. Especially when I'm doing other work as well)
[2002/10/22 08:28:15] #net-snmp <MezoWork> sorry ,)
[2002/10/22 08:28:16] #net-snmp <rstory> wes: i think he means on the agent side...
[2002/10/22 08:28:23] #net-snmp <MezoWork> it should'nt but it does ...
[2002/10/22 08:28:29] #net-snmp <rstory> MezoWork: that's the way snmptable works... it uses get-next
[2002/10/22 08:28:41] #net-snmp <wes> snmptable uses GETBULK, which has horrible overshoot problems and snmptable probably is requesting more data than is there
[2002/10/22 08:28:55] #net-snmp <wes> (getbulk gets translated into getnext-look-alikes inside)
[2002/10/22 08:29:08] #net-snmp <MezoWork> ok. ... so it will be a bug in snmptable ...
[2002/10/22 08:29:11] #net-snmp <rstory> after it (the agent) finishes your table, it looks for the next.. but snmptable shouldn't display any data from other tables
[2002/10/22 08:29:35] #net-snmp <wes> try running snmptable with -CB and it shouldn't walk the entire next table.
[2002/10/22 08:29:36] #net-snmp <MezoWork> snmptable don't show info of other table ... but all are browse
[2002/10/22 08:29:42] #net-snmp <MezoWork> I try
[2002/10/22 08:29:46] #net-snmp <wes> (it'll always request the first variable in the next table)
[2002/10/22 08:30:02] #net-snmp <MezoWork> that's normal ...
--> delanne_ has joined #net-snmp
[2002/10/22 08:40:28] #net-snmp <MezoWork_> sorry ...
[2002/10/22 08:40:33] #net-snmp <MezoWork_> I was disconnected
--> ChanServ has joined #net-snmp
--> wes has joined #net-snmp
--> jbpn has joined #net-snmp
--> wnp22 has joined #net-snmp
--> kary has joined #net-snmp
--> bri99e99 has joined #net-snmp
--> rstory has joined #net-snmp
--> hampa has joined #net-snmp
--> slam has joined #net-snmp
[2002/10/22 08:42:42] #net-snmp <MezoWork_> wes: with snmptable -CB only the requested table is browse
[2002/10/22 08:44:59] #net-snmp <wes> yep.
--> delanne_ has joined #net-snmp
[2002/10/22 08:50:53] #net-snmp <MezoWork> I have another question ... always in table_iterator.c
[2002/10/22 08:50:53] #net-snmp <MezoWork> line 222
[2002/10/22 08:50:53] #net-snmp <MezoWork> * compare the node with previous results
[2002/10/22 08:50:53] #net-snmp <MezoWork> */
[2002/10/22 08:50:53] #net-snmp <MezoWork> if (netsnmp_check_getnext_reply
[2002/10/22 08:50:53] #net-snmp <MezoWork> (requests, coloid, coloid_len, index_search,
[2002/10/22 08:50:55] #net-snmp <MezoWork> &results)) {
[2002/10/22 08:50:57] #net-snmp <MezoWork> if test succes
[2002/10/22 08:50:59] #net-snmp <MezoWork> the result variable is set
[2002/10/22 08:51:01] #net-snmp <MezoWork> ... why there's no goto got_result ?
[2002/10/22 08:51:49] #net-snmp <wes> because it's not necessarily the "right one"
[2002/10/22 08:52:07] #net-snmp <wes> IE, the results have to be returned in lexographical order.
[2002/10/22 08:52:28] #net-snmp <wes> we now know that the current one is the "most correct" but there may be more data in the future which will be "more correct"
[2002/10/22 08:52:36] #net-snmp <wes> welcome to snmp hell
[2002/10/22 08:53:13] #net-snmp <MezoWork> ok thanks
[2002/10/22 08:54:01] #net-snmp <MezoWork> snmp hell ;) that's right !
[2002/10/22 09:03:28] #net-snmp <MezoWork> wes : In netsnmp_iterator_info structure, what's the difference between : free_loop_context_at_end, free_loop_context
[2002/10/22 09:03:41] #net-snmp <MezoWork> I only use free_loop_context_at_end in my code for the moment ...
[2002/10/22 09:04:26] #net-snmp <MezoWork> and have mem leak ... maybe because I don't define free_loop_context and free_data_context
[2002/10/22 09:04:29] #net-snmp <MezoWork> ?
[2002/10/22 09:06:58] #net-snmp <wes> free_loop_context_at_end only gets called once.
[2002/10/22 09:07:04] #net-snmp <wes> (when all is said and done).
[2002/10/22 09:07:19] #net-snmp <wes> use it if you only malloc something in the _first call but not in the _next calls.
[2002/10/22 09:07:35] #net-snmp <wes> use the !at_end one if you malloc something for the loop context in every _next call.
[2002/10/22 09:08:09] #net-snmp <wes> If the pointer you pass back for the data context is malloced in any way, then you must use free_data_context which will be called to free the data_context pointer
[2002/10/22 09:10:36] #net-snmp <MezoWork> ok
[2002/10/22 09:11:34] #net-snmp <rstory> wes: maybe that info could be added as comments in the conf file?
[2002/10/22 09:12:10] #net-snmp <wes> probably.
[2002/10/22 09:12:22] #net-snmp <wes> Actually, yes I'm sure it could be added
[2002/10/22 09:12:29] #net-snmp <wes> the technology is there ;-)
[2002/10/22 09:12:57] #net-snmp <wes> ACTION tries to get the theme song to the 6 million dollar man out of his head.
[2002/10/22 09:13:50] #net-snmp <MezoWork> lol
[2002/10/22 09:13:51] #net-snmp <MezoWork> ;)
[2002/10/22 09:13:55] #net-snmp <rstory> oh great, now it's in my head too...
--> delanne_ has joined #net-snmp
[2002/10/22 09:32:41] #net-snmp <wes> ACTION is away: for a bit (1hr)
[2002/10/22 09:38:21] #net-snmp <delanne_> have a nice day all ...
--> slam has joined #net-snmp
--> miike has joined #net-snmp
[2002/10/22 10:37:15] #net-snmp <wes> ACTION is back
--> benr has joined #net-snmp
--> wes has joined #net-snmp
--> benr has joined #net-snmp
--> wnp22 has joined #net-snmp
--> kary has joined #net-snmp
--> bri99e99 has joined #net-snmp
--> jbpn has joined #net-snmp

Powered by: