[2010/09/01 00:36:50] #net-snmp <hjb> http://www.mail-archive.com/net-snmp-users@lists.sourceforge.net/msg24208.html
[2010/09/01 00:36:52] #net-snmp <hjb> anyone has an idea if that is fixed in 5.5?
[2010/09/01 00:36:55] #net-snmp <hjb> i've got the same problem with 5.5 build with gcc on solaris/sparc
--> cyril_ has joined #net-snmp
--> TomY_ has joined #net-snmp
--> af_ has joined #net-snmp
--> sinesio has joined #net-snmp
[2010/09/01 02:25:17] #net-snmp <sinesio> hi guys, I don't know if anyone can help me here
[2010/09/01 02:26:02] #net-snmp <sinesio> I'm trying to configure SNMP on Solaris, and I have been told that the SNMP agent from Solaris is a modification of net-snmp
[2010/09/01 02:27:01] #net-snmp <sinesio> and probably the things that I want are configured the same way
[2010/09/01 02:27:01] #net-snmp <sinesio> I wanted to configure thresholds for the CPU, Memory and Disk
[2010/09/01 02:27:07] #net-snmp <sinesio> whenever the usage is above a certain level I need to trigger an alarm
[2010/09/01 02:27:11] #net-snmp <sinesio> is this possible?
--> hjb has joined #net-snmp
--> sinesio has joined #net-snmp
[2010/09/01 04:09:25] #net-snmp <sinesio> don't know if someone already answered because I logged out
[2010/09/01 04:10:07] #net-snmp <sinesio> this is what I already tried
[2010/09/01 04:10:07] #net-snmp <sinesio> disk / disk /opt 95% disk /tmp disk /var trap2sink IP_ADDRESS public 162 trapsink IP_ADDRESS public 162 informsink IP_ADDRESS public 162 defaultMonitors yes
--> rstory has joined #net-snmp
--> rstory-work has joined #net-snmp
--> hardaker has joined #net-snmp
[2010/09/01 05:45:41] #net-snmp <hjb> is this channel dead? i've asked three times because of my solaris vfstat problem :-(
[2010/09/01 05:45:50] #net-snmp <hjb> no answer at all
[2010/09/01 06:48:46] #net-snmp <sinesio> you will probably have to wait a little longer
[2010/09/01 07:29:55] #net-snmp <hjb> sinesio: longer then three days?
[2010/09/01 07:30:03] #net-snmp <hjb> that's quite long
[2010/09/01 07:30:23] #net-snmp <sinesio> well
[2010/09/01 07:30:27] #net-snmp <sinesio> what can I say
[2010/09/01 07:30:55] #net-snmp <sinesio> it all depends on the guys availability
[2010/09/01 07:31:20] #net-snmp <rstory> and we don't have a lot of solaris guys...
[2010/09/01 07:32:53] #net-snmp <rstory> i looked for that error message in our code base, and couldn't find it..
[2010/09/01 07:35:17] #net-snmp <hjb> rstory: it probably comes from to lib via perror()
[2010/09/01 07:35:26] #net-snmp <hjb> s/to/the/
[2010/09/01 07:36:23] #net-snmp <hjb> maybe the questions can be more generalized: "anyone with >3T drives running
[2010/09/01 07:36:29] #net-snmp <hjb> net-snmp 5.5 seeing this in the logs when you do a snmpwalk of dskTable"
[2010/09/01 07:36:46] #net-snmp <hjb> without concentrating on solaris
[2010/09/01 07:37:04] #net-snmp <hjb> like: anyone with >3TB driver under linux
[2010/09/01 07:38:27] #net-snmp <hjb> rstory: i think i've found the place the error comes from:
[2010/09/01 07:38:29] #net-snmp <hjb> becker@ulysses:~/net-snmp-5.5 $ find . -name '*.c' | xargs grep "statvfs dev/disk"
[2010/09/01 07:38:31] #net-snmp <hjb> ./agent/mibgroup/ucd-snmp/disk.c: setPerrorstatus("statvfs dev/disk");
[2010/09/01 07:41:13] #net-snmp <rstory> try running the agent with -Ducd-snmp/disk to get some debug info..
[2010/09/01 07:42:48] #net-snmp <rstory> what do you get for grep STATFS net-snmp/include/net-snmp/net-snmp-config.h
[2010/09/01 07:44:49] #net-snmp <hjb> rstory: http://pastebin.com/raw.php?i=yv8wJfpE
[2010/09/01 07:46:11] #net-snmp <hjb> rstory: http://pastebin.com/EJ5VvkUM
[2010/09/01 07:47:15] #net-snmp <rstory> ok.. can you isolate a single query (i.e. a get/getnext, not a walk) that produces the error?
[2010/09/01 07:51:48] #net-snmp <hjb> rstory: i'll try...
[2010/09/01 07:53:43] #net-snmp <hjb> becker@ulysses:~/net-snmp-5.5 $ snmpget -v 2c -c public localhost UCD-SNMP-MIB::dskTotal.3
[2010/09/01 07:53:45] #net-snmp <hjb> UCD-SNMP-MIB::dskTotal.3 = No Such Instance currently exists at this OID
[2010/09/01 07:54:29] #net-snmp <rstory> ok, i was just looking at the patch referenced in the email link you posted..
[2010/09/01 07:54:49] #net-snmp <rstory> what it does is add 64bit object to the mib, since the 32 bit one overflowed..
[2010/09/01 07:55:05] #net-snmp <rstory> it didn't make any changes to the stat calls..
[2010/09/01 07:55:50] #net-snmp <rstory> the statvfs man page has this error code: EOVERFLOW Some values were too large to be represented in the returned struct.
[2010/09/01 07:56:00] #net-snmp <rstory> which is what it looks like you're hitting..
[2010/09/01 07:56:01] #net-snmp <hjb> rstory: so the object now would fit the value, but the wrong stat call is used, right?
[2010/09/01 07:56:22] #net-snmp <rstory> yes, that appears to be the case..
[2010/09/01 07:56:45] #net-snmp <rstory> so if you can come up with the right call, we can work on a patch...
[2010/09/01 07:58:03] #net-snmp <hjb> /* transitional large file interface versions */
[2010/09/01 07:58:08] #net-snmp <hjb> /* transitional large file interface versions */
[2010/09/01 07:58:10] #net-snmp <hjb> /* transitional large file interface versions */
[2010/09/01 07:58:17] #net-snmp <hjb> oh, sorry
[2010/09/01 07:58:25] #net-snmp <hjb> there was a network delay
[2010/09/01 07:58:39] #net-snmp <hjb> #if defined(_LARGEFILE64_SOURCE) && !((_FILE_OFFSET_BITS == 64) && \
[2010/09/01 07:58:40] #net-snmp <hjb> !defined(__PRAGMA_REDEFINE_EXTNAME))
[2010/09/01 07:58:42] #net-snmp <hjb> int statvfs64(const char *_RESTRICT_KYWD, statvfs64_t *_RESTRICT_KYWD);
[2010/09/01 07:58:46] #net-snmp <hjb> i think that's the one to use
[2010/09/01 07:59:05] #net-snmp <hjb> (from /usr/include/sys/statvfs.h)
[2010/09/01 07:59:55] #net-snmp <hjb> but i'm not sure about that note "transitional"
[2010/09/01 08:00:00] #net-snmp <hjb> transitional to what?
[2010/09/01 08:00:36] #net-snmp <rstory> are you up to changing to that call and seeing if it works?
[2010/09/01 08:01:25] #net-snmp <hjb> wait, i've found this: http://pastebin.com/syd6jkNt
[2010/09/01 08:01:51] #net-snmp <hjb> seems the interface hasn't changed but the compilation environment somehow has to be adjusted for large file support
[2010/09/01 08:02:00] #net-snmp <hjb> maybe a cofigure option i'm missing?
[2010/09/01 08:03:11] #net-snmp <hjb> perhaps i just have to give the right #define and everything works :-D
[2010/09/01 08:04:36] #net-snmp <hjb> rstory: i think i'm up to read lfcompile64(5) prior to anything else
[2010/09/01 08:06:37] #net-snmp <hjb> maybe i just have to compile a 64 bit binary, not a 32 bit one
--> SJFriedl has joined #net-snmp
[2010/09/01 08:07:28] #net-snmp <hjb> i'll try that again. i had tried to but only had libopenssl at 32bit and didn't want to compile a 64bit version
[2010/09/01 08:09:50] #net-snmp <hjb> hmm, ok. 32bit apps should use the statvfs64 call while 64bit apps can use traditional statvfs
--> ayecee has joined #net-snmp
[2010/09/01 08:17:41] #net-snmp <ayecee> hi. I'd like to output an snmp walk as html with tables as, well, tables. However, so far I can't see a way to identify a table from the mib tree node types; I don't see an obvious define for this in parse.h. Any suggestions?
[2010/09/01 08:22:06] #net-snmp <ayecee> oh hey. does the "indexes" field in the tree node work for this? i.e. the parent of a node with a non-null index is a table?
[2010/09/01 08:30:50] #net-snmp <rstory> ayecee: does it have to be html? have you seen snmptable? (although that is only one table at a time)
[2010/09/01 08:31:10] #net-snmp <rstory> and of course detecting tables when the mib isn't loaded will be tricky
[2010/09/01 08:35:07] #net-snmp <ayecee> yeah, counting on the mibs being present here. I've looked at snmptable, but it just assumes what you've given it is a table, so far as I can tell.
[2010/09/01 08:35:29] #net-snmp <ayecee> it doesn't have to be html, but whatever output is there will eventually be transformed to html.
[2010/09/01 08:36:15] #net-snmp <ayecee> I'd imagine snmptable is written that way exactly because you can't easily identify a table without a mib.
[2010/09/01 08:40:29] #net-snmp <ayecee> i think i've got it sorted out now.
[2010/09/01 08:55:17] #net-snmp <ayecee> except how to get the value of the index. hmm. the index struct seems to only have the label of the index.
[2010/09/01 08:56:32] #net-snmp <ayecee> i guess that would be in the data from the walk, not the mib.
[2010/09/01 09:09:35] #net-snmp <ayecee> not sure how to pull that out. how would I determine that, say, 1.3.6.1.2.1.1.9.1.2.5.0 was the 2nd scalar in the 5th row of sysORTable?
[2010/09/01 09:10:20] #net-snmp <ayecee> I could just grab the second-last number, but that would only work for integer indexes.
[2010/09/01 09:20:03] #net-snmp <ayecee> oooh, parse_oid_indexes. bingo.
[2010/09/01 09:22:05] #net-snmp <ayecee> maybe.
--> hjb_ has joined #net-snmp
--> rstory has joined #net-snmp
--> hesco has joined #net-snmp
[2010/09/01 16:13:41] #net-snmp <hesco> I'm using net-snmp to help opennms opennms monitor a server we're running, but we're not getting any details on the cpu loads. On most of our network we installed from yum, but on this server (which we are not quite ready to migrate from) we have a broken glibc and were left to install from source. Apparently we left something out of our configuration, which otherwise seems to be working. Does this suggest to anyone what I might be mis
[2010/09/01 16:13:41] #net-snmp <hesco> Can anyone point me to a troubleshooting guide which might help me narrow this down?
[2010/09/01 16:20:11] #net-snmp <ayecee> hesco: what did you use for a configure line?
--> rstory has joined #net-snmp
[2010/09/01 16:42:34] #net-snmp <hesco> how do mean ?
[2010/09/01 16:42:49] #net-snmp <hesco> I configured a conf file, not just a line
[2010/09/01 16:43:47] #net-snmp <hesco> included was this line: view systemview included .1
[2010/09/01 16:44:45] #net-snmp <hesco> in fact, the entire config file is here: http://pastebin.com/uCsqKU5c
--> rstory has joined #net-snmp
--> rstory has joined #net-snmp
--> Linkadmin has joined #net-snmp
--> krphop has joined #net-snmp
--> tris has joined #net-snmp
--> snmpbot has joined #net-snmp