I use net-snmp-5.4 on linux-2.6.16 with sk98lin driver.
If link of some interface is off, snmpd will core dump periodically.
After investigation, I find that netsnmp_linux_interface_get_if_speed
will cause stack corrupt.
If link is off, speed returned by ETHTOOL_GSET will be 0(not 10 100 1000).
A
I use net-snmp-5.4 on linux-2.6.16 with sk98lin driver.
If link of some interface is off, snmpd will core dump periodically.
After investigation, I find that netsnmp_linux_interface_get_if_speed will
cause stack corrupt.
If link is off, speed returned by ETHTOOL_GSET will be 0(not 10 100 1000).
Glenn wrote:
> The fix is to change table->next_index to table->next_index-1. We've
> tried this fix locally and it definitely solved our problems. I've
> submitted patch 1752934 with this fix.
Much appreciated. We'll look at applying that patch once 5.4.1 is out of
the door.
+Thomas
We've had this intermittent bug in our of our custom MIB implementations
that has been driving us nuts for some time. We just recently managed
to get valgrind on the embedded device and we discovered a problem in
the net-snmp library.
The MIB in question was written some time ago and it makes
Hi!
It is Basic question in snmp. when two snmp managers trying to
set a variable on same snmp agent at the same time. how request will be
processed.
I am using latest net-snmp source code that is
net-snmp5.3.1. in this code i couldn't found.
If any body kno
Got some rather large cores (4gb) from one of our net-snmp 5.3 boxes.
Apparently they have been doing this randomly. I have at least 4 core files
and
I have pulled the backtrace on the core file and was wondering if
anyone has any ideas.Core was generated by `snmpd'.Program terminated with
Robert here's the bug url.
http://sourceforge.net/tracker/index.php?func=detail&aid=1752162&group_id=12694&atid=112694
looping in
asn_realloc_rbuild_string using up 90+% of CPU and eating up memory at
about 1 Meg every 5 seconds
backtrace for the hang is:
#0 0x8c10 in __memcpy ()
#1 0x0017ae0
I noticed the following:
a) when the subagent is being compiled, it will look for the library files in
the "/usr/local/lib" directory instead of the
"/export/home/rosent1/snmp/usr/local/lib" directory where the library files
were created. Basically, I must copy the library files into the
"/u
As for (2) below, I realized I recently deleted all the library files from the
"/usr/local/lib/" directory from existing since I thought they were not being
used since the library files which should be used by the subagent were the
files located in the "/export/home/rosent1/snmp/usr/local/lib/"
1) What do you mean by "You can copy the prototypes from the appropriate
headers, if you want to eliminate the warning". What do I copy from where? Do
you mean to simply add the #include files to the subagent.c file?
2) The following library files are located in the
"/export/home/rosent1/snm
On Thu, 12 Jul 2007 08:28:25 -0700 (PDT) Need wrote:
NH> gcc -I. `net-snmp-config --base-cflags` -Wall -Wstrict-prototypes -c -o
ocStbHostMib_subagent.o ocStbHostMib_subagent.c
NH> ocStbHostMib_subagent.c: In function `main':
NH> ocStbHostMib_subagent.c:76: warning: implicit declaration of funct
On Thu, 12 Jul 2007 07:49:11 -0700 (PDT) Need wrote:
NH> Anyway, my question is, once a new officially supported NetSNMP version is
released then must I regenerate all of the code using the "mib2c.mfd.conf"
configuration file? Or perhaps the MFD code did not change in this next
release and I wi
I have progressed but have another problem.
I used mib2c with the MFD configuration file to generate code for three tables
(ocStbHostAVInterfaceTable, ocStbHostAnalogVideoTable and
ocStbHostComponentTable) and one scalar (ocStbHostHWIdentifiers). I am
compiling on my PC (ie: not cross-compilin
I need to upgrade to the next NetSNMP version when it is released since it has
an important "../agent/mibgroup/Makefile" change which will allow all of the
".o" and ".lo" files to be deleted when a "make clean" requested is issued.
The current v5.4 version does not handle "make clean" correctly
Ok, I think I figured this one out. Seems like the library files were not
being produced since the agent/mibgroup ".o" and ".lo" files were not being
removed due to the make clean request. I fixed this error in the Makefile one
time but have since performed a "./configure" again which wiped
BAUDRY Charles wrote:
> /usr/bin/ld: fsprpc_clnt.a(fsprpc_c.o): Relocations in generic ELF (EM: 20)
> fsprpc_clnt.a: could not read symbols: File in wrong format
> collect2: ld returned 1 exit status
> make: *** [snmpd] Error 1
We don't know anything about your fsprpc_clnt.a and/or your platform,
Hello
I tried to add my library with the options --with-ldflags=-L/path/to/dir
--with-libs=fsprpc_clnt.a
It didn't work and the make step failed, it says that it doesn't find the
file. As said in the previous message, the ./configure find the file but not
the make...
I searched in the mailing lis
17 matches
Mail list logo
