VIMAGE: Freed UMA keg was not empty

[Bjoern A. Zeeb]

On Fri, 14 Jan 2011, Thierry Herbelot wrote:

> "Bjoern A. Zeeb" <bzeeb-lists at lists.zabbadoz.net> a écrit
>> On Thu, 18 Nov 2010, Thierry Herbelot wrote:
>>> "Bjoern A. Zeeb" <bzeeb-lists at lists.zabbadoz.net> a écrit
>>>
>>>> On Thu, 18 Nov 2010, Thierry Herbelot wrote:
>>>>> "Bjoern A. Zeeb" <bzeeb-lists at lists.zabbadoz.net> a écrit
>>>>>
>>>>>> On Wed, 17 Nov 2010, Thierry Herbelot wrote:
>>>>>>> As promised, here are the full logs (in attachment)
>>>>>>>
>>>>>>> This is a serial console log showing the command loop that triggers
>>>>>>> the bug on a debug kernel and ensuing DDB session.
>>>>>>>
>>>>>>> the obvious problem line is :
>>>>>>>          routetbl         2684       303890K         3469
>>>>>>>
>>>>>>> (further tests showed an increase of the routetbl malloc zone by
>>>>>>> 4MBytes for each vnet jail creation/destruction cycle)
>>>>>>
>>>>>> Hmm, I had fixed that (somewhere).  I'll see where the patch went. You
>>>>>> are on 8.1-RELEASE or -STABLE?
>>>>>
>>>>> This will be for a -release, thus 8.1 for now, but we will switch to
>>>>> 8.2 ASAP
>>>>
>>>> Well wait; I am not sure the changes are in SVN at all.  I'll get back
>>>> to you.
>>>
>>> OK for us : we are still in the process of deploying the solution, but
>>> still we will test any patch you can forward ;-)
>>
>> Do you have FLOWTABLE enabled in your kernel config?
>
> Hello,
>
> I will check at work on monday, but I think so, as the kernel configuration
> was derived from GENERIC.

that would have been the reason for the memory leaks.


> BTW, the setup we wanted (ie with VIMAGE) is showing good stability (with 8.2-
> stable) as we have removed the need to start and stop the jails. we will

but with that you won't see it anymore.

/bz

-- 
Bjoern A. Zeeb                                 You have to have visions!
         <ks> Going to jail sucks -- <bz> All my daemons like it!
   http://www.freebsd.org/doc/en_US.ISO8859-1/books/handbook/jails.html