On Sun, May 31, 2009 at 09:03:23AM +0200, Thomas Anders wrote:
> John Morrissey wrote:
> > Just tried this; snmpd does *not* leak when kernel IPv6 support is
> > absent.
>
> Thanks for your great help so far. Do you also have a chance to test
> upstream's 5.4.x SVN and/or SVN trunk (with IPv6 supp
John Morrissey wrote:
> Just tried this; snmpd does *not* leak when kernel IPv6 support is absent.
Thanks for your great help so far. Do you also have a chance to test upstream's
5.4.x SVN and/or SVN trunk (with IPv6 support enabled again)?
+Thomas
--
To UNSUBSCRIBE, email to debian-bugs-di
On Fri, May 29, 2009 at 08:53:08PM +0200, Jochen Friedrich wrote:
> Do you have a chance to test the same on a system without IPv6
> support? I still suspect the leak when trying to add duplicate (link
> local) IPv6 addresses to the interface table.
Just tried this; snmpd does *not* leak when kern
Hi John,
> 5.4.2.1 leaks in the same way/amount as 5.4.1~dfsg-12. After ~3.5 hours of
> uptime:
Do you have a chance to test the same on a system without IPv6
support? I still suspect the leak when trying to add duplicate (link
local) IPv6 addresses to the interface table.
Thanks,
Jochen
--
On Fri, May 29, 2009 at 02:42:25PM +, John Morrissey wrote:
> This version of net-snmp is also in sid, so I'm currently trying the
> latest upstream release (5.4.2.1) to see if it's any different.
5.4.2.1 leaks in the same way/amount as 5.4.1~dfsg-12. After ~3.5 hours of
uptime:
==25243== 1,4
Package: snmpd
Version: 5.4.1~dfsg-12
Severity: normal
lenny's snmpd leaks memory when VLAN subinterfaces are present. For example,
machines here with 13 VLAN subints leak about 200mbytes of memory every
couple of weeks. Other amd64 machines without VLAN subints do not leak.
The observed behavior
6 matches
Mail list logo
