snmpusm for
modifying the passwords of an existing snmp user,
fails with Not in time window authentication error.
What exact version are you working with? This issue should be fixed in
any recent Net-SNMP release.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
.
Now that a proposed patch has been uploaded to
http://sf.net/tracker/index.php?func=detailaid=1641685group_id=12694atid=312694
could you please comment on whether/how the proposed patch deals with
the open issues mentioned above?
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
for all versions (Tested on kernel
2.6.18.6, if someone could test it on older versions as well then that
would be nice)
Committed to 5.[234].x CVS and CVS MAIN. Thanks for the patch.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
Dave Shield wrote:
On 15/01/07, Thomas Anders [EMAIL PROTECTED] wrote:
thanks for the Wiki feedback. Do you care to propose a specific patch
what
*exactly* you'd like to drop w.r.t. the third issue?
Appended.
I don't think this patch is complete or would build as-is. Don't we need
Dave Shield wrote:
On 18/01/07, Thomas Anders [EMAIL PROTECTED] wrote:
I don't think this patch is complete or would build as-is. Don't we need
the attached one at least?
Probably, yes.
[...]
1) Can you post your proposed patch for the dropauth/SyslogTrap
integers, please?
If you
to conditionally call the helper's init function if a
static need_init flag is set.
Interesting idea, as it wouldn't affect backwards compatibility, AFAICS.
Could any of the other developers with major helper experience comment
on the feasibility of this approach, please?
+Thomas
--
Thomas Anders
hr_network.c build failure on OS X
http://sourceforge.net/tracker/index.php?func=detailaid=1627739group_id=12694atid=112694
See the bug tracker item for suggested workarounds.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
Dave Shield wrote:
On 15/01/07, Thomas Anders [EMAIL PROTECTED] wrote:
thanks for the Wiki feedback. Do you care to propose a specific patch
what
*exactly* you'd like to drop w.r.t. the third issue?
Appended.
Fine by me, especially since I have no idea what this code was ever
used/useful
Robert Story wrote:
On Thu, 11 Jan 2007 13:44:16 -0800 Wes wrote:
WH TA == Thomas Anders [EMAIL PROTECTED] writes:
WH
WH TA Dave Shield wrote:
WH The distinction between libagent and libhelpers doesn't feel as useful
WH as that between these two and libmibs
WH
WH TA Merging them
tried recently), please let us know.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share
Thomas Anders wrote:
Starting with the 5.4 release cycle there have been some (more) spurious
discussions about library layering and dependencies. I've been trying to
summarize the issue(s) here:
http://www.net-snmp.org/wiki/index.php/Library_layering
I've updated that page to reflect
-I/usr/lib/perl5/5.8.8/i386-linux-thread-multi/CORE conftest.c 5
In file included from conftest.c:134:
/usr/include/linux/ethtool.h:18: error: expected
specifier-qualifier-list before '__u32'
net-snmp version? OS version? configure options?
+Thomas
--
Thomas Anders (thomas.anders at blue
Wes Hardaker wrote:
TA == Thomas Anders [EMAIL PROTECTED] writes:
TA How much of the snmpv3/usmUser core code could likely be reused in a new
TA implementation with one of those two helpers?
Probably just the case statement that figures out the right values for
a given column
Dave Shield wrote:
On 08/01/07, Thomas Anders [EMAIL PROTECTED] wrote:
How about the attached patch?
Looks good to me.
Checked in (with a necessary include added).
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
Alef Veld wrote:
In that case, does anyone have any good example code for a
snmpdtrapd ?
See apps/snmptrapd.c in the Net-SNMP source.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn
with the
critical dependencies listed in the table there.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get
this is undefined,
and make sure that HAVE_KVM_H *is* defined.
But this may well be wrong,incomplete,etc
I'd rather do this change to include/net-snmp/system/solaris.h, in fact,
but haven't tested it myself, so YMMV.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT business topics through brief surveys
Magnus Fromreide wrote:
The attached patch makes it possible to build snmplib/ with a c++
compiler as well as a c89 compiler.
Much appreciated. I'll check it in, unless specific objections are raised.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
. Unfortunately this breaks the build with g++ (4.1.0) for me:
snmp_api.c:768: error: 'netsnmp_register_service_handlers' was not declared in
this scope
Comments/patches appreciated.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
Thomas Anders wrote:
Thanks for the quick response. Committed.
However, g++ (4.1.0) still doesn't seem to like snmp_service.c too much:
snmp_service.c: In function 'int netsnmp_register_default_domain(const char*,
const char*)':
snmp_service.c:35: error: invalid conversion from 'void
Dave Shield wrote:
If you try
cat /proc/05026/psinfo
how much does this report? Everything, or is this truncated too?
None of that, just binary data, I fear. :-)
Unlike Linux /proc, Solaris /proc is designed for C, not cat.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
, but it could be changed to
aType* foo = (aType*)malloc(sizeof(aType));
to move it into the common subset of C and C++.
As most of the code already builds fine with g++, /me thinks it should
be done.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
been dropped in the new? I can't believe there's no good way
to get NETSNMP_MEM_TYPE_VIRTMEM values on Solaris.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT business topics through brief surveys
-2.61 make make install
ln -s ../autoconf-2.61/bin/autoconf /usr/local/bin/autoconf2.61
ln -s ../autoconf-2.61/bin/autoheader /usr/local/bin/autoheader2.61
Then just call autoconf2.61 (or autoheader2.61, respectively) whenever
you need to run autoconf/autoheader for MAIN.
+Thomas
--
Thomas Anders
Siva Prakash Reddy G wrote:
I'd like to prevent messages like 'Connection from UDP: ip'
from popping up in the logfiles i.e., in syslog file, after running the
agent.
What platform and net-snmp package are you working with?
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
, unpack the source and run ./configure
... make make install.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get
Dave Shield wrote:
On 03/01/07, Thomas Anders [EMAIL PROTECTED] wrote:
Please start again from scratch, unpack the source and run ./configure
... make make install.
Or 'make distclean ; ./configure '
should have the same effect.
Only if you ran make distclean *before* you run
Thijs Brouwer wrote:
Is there already someone working on this? If not, what would be the
procedure to get this fix into the official distribution?
Please post it to the project's patches tracker
(http://www.net-snmp.org/patches) so it won't get lost. Thanks.
+Thomas
--
Thomas Anders
2.61 there.
There's also the datarootdir problem when upgrading autoconf:
http://groups.google.com/group/mailing.unix.net-snmp-coders/browse_thread/thread/83954dbd7d771655/fda81537da568e08
so I'm not sure yet what to do with the branches.
+Thomas
--
Thomas Anders (thomas.anders at blue
Magnus Fromreide wrote:
On Tue, Dec 12, 2006 at 12:23:15AM +0100, Thomas Anders wrote:
are there any objections to update to current stable autoconf (2.61) for CVS
MAIN?
Excellent idea! I am all in favour of it.
Here you are. configure in MAIN is now generated with 2.61. It doesn't
use any
-snmp.cvs.sourceforge.net/net-snmp/net-snmp/agent/mibgroup/host/hr_swrun.c?r1=5.8.2.5r2=5.8.2.6
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's
isn't a bad idea either.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions
Magnus Fromreide wrote:
The patch made snmptrap run successfully but it generates the message
Creating directory: /var/net-snmp/mib_indexes
even when not doing so.
Should now be fixed, too.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
Win32 handles remain open.
Do we have a policy whether to apply Win32-centric NETSNMP_* defines to
win32/net-snmp-config.h[.in] only (like done here) or to the
include/net-snmp/net-snmp-config.h.in master copy also? If not, what shall the
policy look like? :-)
+Thomas
--
Thomas Anders
are the other
practical options?
Suggestions highly appreciated.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get
Dave Shield wrote:
On 22/12/06, Thomas Anders [EMAIL PROTECTED] wrote:
I realize that snmpv3/usmUser is v4/ucd-based. What are the options for
performance improvements?
My gut reaction would be to re-implement this module using one of the
v5-handlers - perhaps tweaking the snmplib/snmpusm.c
() implementations
used on Solaris vs. Linux, as discussed on the lists before.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll
Wes Hardaker wrote:
GSM == G S Marzot [EMAIL PROTECTED] writes:
GSM thoughts?
Have you read our previous discussions on the subject?
Do you have a pointer to it? I vaguely remember this may have just been
some IRC talk between the two of us. I hope I'm wrong.
+Thomas
--
Thomas Anders
Wes Hardaker wrote:
TA == Thomas Anders [EMAIL PROTECTED] writes:
TA are there any objections to update to current stable autoconf
TA (2.61) for CVS MAIN?
What are the advantages of requiring it?
(as opposed to simply using a fairly recent one)
I remember quite a few occasions where I
building (and only very few systems already ship 2.61). OTOH, it may
make life easier for core developers. I'd propose to update MAIN first
and see how it goes.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
code doing similar things and relying
on our backwards-compat promise would hit the same issue, wouldn't it?
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join
This is known bug #1588455.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions
in *addition* to 2.59.
I'd like to do it rather early than later in the 5.5 development cycle to reduce
the risk and give us enough time to clean up configure.in according to current
best practice.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
-Perl Net-SNMP config.
What's the output of:
- net-snmp-config --configure-options
- net-snmp-config --agent-libs
? Is this the *only* Net-SNMP installation on this box?
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
://www.net-snmp.org/bugs, please?
Also, it may be worth looking through the compiler warnings for a 64-bit
build. These are (way too) many, unfortunately, but it'd be a start at
least and patches to squelch them would be highly appreciated.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
=12694
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT business topics
about the reason of this change. The corresponding commit log message clean up
compile and fix tests isn't too helpful, either. :-/
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence
--with-logfile=/var/log/snmpd.log --with-persistent-directory=/var/net-snmp
--with-default-snmp-version=2 --with-out-perl-modules
--disable-embedded-perl --prefix=Mydir_PATH/install
Like I said:
Thomas Anders [EMAIL PROTECTED] wrote:
... --without-perl-modules ..., actually.
+Thomas
in README.aix.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions
control.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT business
, but not
specifically determinant as yet.
I noticed 5.4 has been flagged for supported, is it worth upgrading to 5.4
to fix this issue?
I'd try running 5.4 for =1 week before anything else.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
Dana Burns wrote:
Magnus Fromreide wrote:
What can apps/snmpcomm do that snmpset can't do if we assume that your
mib module is available?
Probably nothing other than conveniently packing the needed snmpsets
together?
I could certainly see the benefit of that.
+Thomas
--
Thomas Anders
determination
Please let us know whether/how it works for you.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll
or directory
Looks like your gcc installation is broken^Wincomplete.
Where *is* libgcc.so.1 on your system? You need to add this directory to
your crle config.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take
for taking the time to do the updating and sharing it with us.
5.4 ships with the updated README.smux.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join
raises late objections or proposes modifications.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance
.
And did you test non-64bit again, to make
sure it didn't break anything?
Sure, I've said tested on RHEL4/x86 and that implied *with* the patch,
of course.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take
Dave Shield wrote:
I'm slightly surprised the find uses a wildcard rather than the _libdir
macro,
but if it works, then fair enough.
find uses usr/lib... (*without* a leading slash), so _libdir wouldn't work.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
. (real memory) and (swap space or
virtual memory).
Have you run the final MIB (after the patch) through smilint?
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future
I've just updated dist/net-snmp.spec in CVS MAIN.
I've tested it on Fedora 5 i586 and it works for me.
Is there anyone willing to test it on other Fedora versions (FC6 preferred),
architectures (x86_64) and/or --without embedded_perl?
Feedback appreciated.
+Thomas
--
Thomas Anders
Wes Hardaker wrote:
TA == Thomas Anders [EMAIL PROTECTED] writes:
TA Given that 5.4 enables embedded Perl and Perl modules by default,
shouldn't
TA dist/net-snmp.spec be updated to reflect that?
TA Otherwise, I'm not sure a default rpmbuild will work too well.
Probably, but I think
pushing rc3. :-)
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions
Wes Hardaker wrote:
TA == Thomas Anders [EMAIL PROTECTED] writes:
TA snmplib/system.c uses nlist() (and open(/dev/kmem), lseek and read)
TA without properly protecting the code by #ifdef NETSNMP_CAN_USE_NLIST.
TA Thus, this code is active even if --without-kmem-usage has been
TA specified
try the MSVC Workspace build on Windows for
increased confidence?
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel
of legal options! Patch #1598968, submitted by Peter Eckel and
attached here, fixes this. I've applied it to the 5.3.x branch and I'm calling
for votes to also fix this for 5.4.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
diff -ur net-snmp-5.4.rc3/apps/snmptrapd.c net-snmp-5.4.rc3
tools we
don't AC_PROG()-check for?
Before I go and try to understand Wes' related recent configure changes: what
shall I check specifically in order to track this down?
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
and it works fine
for me.
I'm calling for [quick] votes on whether/when to apply this patch for 5.4.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
Index: agent/mibgroup/ucd-snmp/memory.c
===
RCS file: /cvsroot/net-snmp/net
Given that 5.4 enables embedded Perl and Perl modules by default, shouldn't
dist/net-snmp.spec be updated to reflect that?
Otherwise, I'm not sure a default rpmbuild will work too well.
@Wes: Do spec file changes require a rfv?
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
G. S. Marzot wrote:
this fix touches no other previously released Net-SNMP feature so should be
innocuous to accept.
Fine by me for this reason and because it defaults to off.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
I remember is this patch:
http://sourceforge.net/tracker/index.php?func=detailaid=1420758group_id=12694atid=456380
But I'd have thought this particular issue was restricted to 5.3/5.3.0.1.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
-RELEASE-p9 FreeBSD 5.4-RELEASE-p9 #0: Thu Jan 12 05:31:33 UTC 2006
[EMAIL PROTECTED]:/usr/obj/usr/src/sys/GENERIC i386
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Take Surveys. Earn Cash. Influence the Future
Dave Shield wrote:
On 13/11/06, Thomas Anders [EMAIL PROTECTED] wrote:
I'm not sure I see the benefits of tweaking the MIB and the code to
replace virtual memory by real+swap for these objects.
The changes to the MIB are intended to make the behaviour more
explicit - thus avoiding
.
- UCD memory reporting (memFree swapError)
so far purely a debate between me and Thomas :-(
Too true.
- perl cc detection
RFV posted. No single feedback yet, so at +1.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
build process, thereby fixing bug 1577913 raised by Apple. I'm calling for
votes to include this in 5.4.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
Index: man/Makefile.in
===
RCS file: /cvsroot/net-snmp/net-snmp/man
G. S. Marzot wrote:
+1 - I will be happy to review the changes too.
They're already in CVS MAIN. Feel free to double-check them.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Using Tomcat but need to do
-280
(W) Function argument assignment between types void(*)(struct interpreter*)
and void(*)(void) is not allowed.
gmake[1]: *** [snmp_perl.lo] Error 1
gmake[1]: Leaving directory `/home/anders/build/5.4cvsaix/yawara/agent'
Thoughts?
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
with the getopt behaviour on Solaris.
IIRC --token=value doesn't work either. Patches appreciated.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Using Tomcat but need to do more? Need to support web services
+real). I recognise that any change of behaviour is
unwelcome, but I suggest that the benefits of a clear and consistent
approach outweigh any disadvantages.
Fine by me.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
,
because I find it easier to read. YMMV, of course.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
Index: snmplib/system.c
===
RCS file: /cvsroot/net-snmp/net-snmp/snmplib/system.c,v
retrieving revision 5.33
diff -u -p -r5.33
nlist symbol in libnetsnmp which at least breaks the perl
modules. I don't have the exact error message handy, but I can say that
my patch fixes the issue there.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
Index: perl/SNMP/SNMP.xs
===
RCS file: /cvsroot/net-snmp/net-snmp/perl/SNMP/SNMP.xs,v
retrieving revision 5.69
diff -u -p -r5.69 SNMP.xs
--- perl/SNMP/SNMP.xs 25 Oct 2006 16
Thomas Anders wrote:
t/mib.dubious
Here's the offending call that fails on 64-bit:
SNMP::translateObj('.iso.org.dod.internet.mgmt.mib-2.system.sysDescr');
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
thought?
Not impossible, actually, given that we don't have a dedicated MIB
handling test in our main test suite, but only the mentioned SNMP.pm
t/mib.t in make perltest.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
(including subsequent make after no
changes or make test or whatever) which sucks. Can we fix this for
rc2, please?
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Using Tomcat but need to do more? Need
Dave Shield wrote:
On 07/11/06, Thomas Anders [EMAIL PROTECTED] wrote:
On what systems has this been tested?
FreeBSD - can't remember which version offhand.
(I'd need to reboot to check). But cpu_nlist is only
active on FreeBSD kit anyway (see hardware/cpu.h)
Note that this fix matches
.
Anyone else seeing this? Ideas? Fixes?
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology
Wes Hardaker wrote:
TA == Thomas Anders [EMAIL PROTECTED] writes:
TA Building CVS MAIN on Linux, darix and I noticed that make
TA perltest fails in the SNMP.pm MIB code tests:
I'm actually not sure that the SNMP module has worked on 64 bit
systems recently at all. Or at least we keep
Dave Shield wrote:
On 03/11/06, Thomas Anders [EMAIL PROTECTED] wrote:
testing 5.4.rc1 on the SF Compile Farm went pretty well, *except* for
x86-freebsd1 (FreeBSD 5.4). make test fails miserably on a large
number of tests.
I've investigated this, and it turns out that the problem is down
Magnus Fromreide wrote:
On lör, 2006-11-04 at 15:09 +0100, Thomas Anders wrote:
(i.e. proxy over TCP) will fail. UDP works fine, as well as starting the
proxy
destination agent before the agent itself.
Hmm. To me it looks like the problem is that we aren't considering
connection oriented
) --without-kmem-usage with the patch now
prevents the agent from hanging when walking ifTable which is what the
default build suffers from.
I'm calling for votes to include this patch in 5.4.rc2. I'll update
documentation with notes on --without-kmem-usage once the patch is in.
+Thomas
--
Thomas Anders
for the confusion.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
Index: agent/kernel.h
===
RCS file: /cvsroot/net-snmp/net-snmp/agent/kernel.h,v
retrieving revision 5.6
diff -u -p -r5.6 kernel.h
--- agent/kernel.h 31 Oct
modules feels like much more
risk (to me at least) than the straightforward job of adding error
handling the way I've done.
But I'll be trying to implement whatever the consensus is, as long as
we're not keeping things as broken as they are.
Comments highly appreciated.
+Thomas
--
Thomas Anders
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere
module list is *much*
larger than the list of modules that use nlist/klookup at all. E.g. AIX 5.x has
full working host mib support, even on 64-bit, so forcing them to use
--enable-mini-agent will drop a lot.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
Robert Story wrote:
Give that Thomas tested w/out issues: +1
Is there any reason why this hasn't been applied yet even though we're at +3?
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Using Tomcat
?
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de)
-
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM
in init_mib_modules () at mib_module_inits.h:89
#3 0x0804c577 in main (argc=2, argv=0xbfbfeb04)
at /home/users/t/ta/tanders/src/net-snmp-5.4cvs/agent/snmpd.c:808
Hopefully we'll be able to fix this before 5.4 final.
+Thomas
--
Thomas Anders (thomas.anders at blue-cable.de
G. S. Marzot wrote:
Attached is a patch which addresses a few memory leak issues in the Python
Net-SNMP interface.
The proposal is that these fixes be accepted prior to release.
+1 (especially since the Python modules haven't been shipped before)
+Thomas
--
Thomas Anders (thomas.anders
401 - 500 of 1040 matches
Mail list logo