Bug#500717: [Pkg-net-snmp-devel] Bug#500717: snmpd reports assert in syslog and does not return data

2008-10-04 Thread Jochen Friedrich
Hi Thomas,

 Anders schrieb:
 Jan Evert van Grootheest wrote:
 Since upgrading from etch (stable) to testing, snmpd is reporting this in 
 syslog:
 Sep 30 06:25:22 quark snmpd[4507]: netsnmp_assert rc == 0 failed 
 if-mib/ifTable/ifTable_data_access.c:210 _check_interface_entry_for_updates()
 [...]
 If you think the libc6 or kernel below looks strange, that's because this is 
 a xen dom0 and it uses
 
 This is a known bug, only triggered with Xen, with a known fix. See
 upstream changeset  r16790:
 
 http://net-snmp.svn.sourceforge.net/viewvc/net-snmp?view=revrevision=16790
 
 This change probably needs to be applied to the lenny package to fix the
 problem in question.

Yes. I'll include this in my next (and hopefully last) update for Lenny.

Thanks,
Jochen



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Bug#500717: [Pkg-net-snmp-devel] Bug#500717: snmpd reports assert in syslog and does not return data

2008-10-02 Thread Jan Evert van Grootheest

Jochen,


Since upgrading from etch (stable) to testing, snmpd is reporting this in 
syslog:
Sep 30 06:25:22 quark snmpd[4507]: netsnmp_assert rc == 0 failed 
if-mib/ifTable/ifTable_data_access.c:210 _check_interface_entry_for_updates()



Does this still happen, if you restart snmpd? It looks like this might happen 
if the interface name is changed after snmpd has been started.
  

At restart this happened:
Oct  2 16:38:11 quark snmpd[4720]: netsnmp_assert rc == 0 failed 
if-mib/ifTable/ifTable_data_access.c:210 
_check_interface_entry_for_updates()
Oct  2 16:42:15 quark snmpd[4720]: Received TERM or STOP signal...  
shutting down...
Oct  2 16:42:17 quark snmpd[4033]: netsnmp_assert !registration != 
duplicate failed agent_registry.c:535 netsnmp_subtree_load()

Oct  2 16:42:17 quark snmpd[4033]: NET-SNMP version 5.4.1

But it does no longer repeat the assert. Now that you mention this, I 
have to report that previously (i.e. etch snmpd) would crash at reboot.



If you think the libc6 or kernel below looks strange, that's because this is a 
xen dom0 and it uses
debian testing now, except for xen and the kernel. That's from ubuntu. But xen 
requires libc6 2.8
(it is the latest; I installed that today). And I use that kernel for HW 
support.
(now that I'm at it, debian is not intending to ship 2.6.18 for xen in lenny, 
does it?)



Are you using the standard Debian startup scripts or something else?
  
This is the full list of ubuntu packages: findutils, libc6, libc6-i386, 
libxen3, python-xen-3.3, xen-hypervisor-3.3, xen-utils-3.3.
In reality this is because I want to run xen 3.3, which requires the 
newer libc, libxen3 and python-xen-3.3.


But I'm just using the regular startup scripts that came with the packages.

Did I already mention that this is a xen dom0? I.e. after startup, 
network interfaces appear to connect to the domU instances, the physical 
ethernet gets renamed and a bridge gets added. So perhaps that's a bit 
complex for snmpd? I don't know.


If there's something I can do to help, like get you a crash dump or a 
backtrace with gdb , please let me know (I'm a programmer myself, just 
no experience with snmpd).


Thanks,
Jan Evert





--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Bug#500717: [Pkg-net-snmp-devel] Bug#500717: snmpd reports assert in syslog and does not return data

2008-10-01 Thread Jochen Friedrich
severity 500717 important
thanks

Hi Jan,

 Severity: grave
 Justification: renders package unusable

Changed to important because the bug does NOT render the package unusable to 
everyone.

 Since upgrading from etch (stable) to testing, snmpd is reporting this in 
 syslog:
 Sep 30 06:25:22 quark snmpd[4507]: netsnmp_assert rc == 0 failed 
 if-mib/ifTable/ifTable_data_access.c:210 _check_interface_entry_for_updates()

Does this still happen, if you restart snmpd? It looks like this might happen 
if the interface name is changed after snmpd has been started.

 If you think the libc6 or kernel below looks strange, that's because this is 
 a xen dom0 and it uses
 debian testing now, except for xen and the kernel. That's from ubuntu. But 
 xen requires libc6 2.8
 (it is the latest; I installed that today). And I use that kernel for HW 
 support.
 (now that I'm at it, debian is not intending to ship 2.6.18 for xen in lenny, 
 does it?)

Are you using the standard Debian startup scripts or something else?

Thanks,
Jochen



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]