Winbindd using 100% of CPU. Any solution?

[Richard Sharpe]

On Wed, Dec 18, 2013 at 10:14 AM, Jeremy Allison <jra at samba.org> wrote:
> On Tue, Dec 17, 2013 at 03:24:15PM -0800, Richard Sharpe wrote:
>> On Mon, Dec 16, 2013 at 1:56 PM, Richard Sharpe
>> <realrichardsharpe at gmail.com> wrote:
>> > On Mon, Dec 16, 2013 at 11:10 AM, Richard Sharpe
>> > <realrichardsharpe at gmail.com> wrote:
>> >
>> > More info ... because I see this in the log:
>> >
>> >    adding 0x80334c940 to list at 0xeac360
>> >
>> > but do not see this:
>> >
>> >    removing %p from list %p
>> >
>> > and because I see that the PID has been initialized, it looks like
>> > child->sock has been set to -1 without removing the child from the
>> > winbindd_children list!
>>
>> OK, I think I have the smoking gun.
>>
>> Here is what I see in the log:
>>
>> [2013/12/17 16:34:08.459496, 10]
>> winbindd/winbindd_dual.c:1385(fork_domain_child)
>>   fork_domain_child called for domain 'EXCHANGE'
>> [2013/12/17 16:34:08.460496, 10]
>> winbindd/winbindd_dual.c:1438(fork_domain_child)
>>   Child process 15002
>> [2013/12/17 16:34:08.460496,  0]
>> winbindd/winbindd_dual.c:1430(fork_domain_child)
>>   adding 0x803358760 to list at 0x80330b100
>>
>> then a little while later:
>>
>> [2013/12/17 16:39:07.843301, 10]
>> winbindd/winbindd_dual.c:1385(fork_domain_child)
>>   fork_domain_child called for domain 'XCHANGE'
>> [2013/12/17 16:39:07.844302, 10]
>> winbindd/winbindd_dual.c:1438(fork_domain_child)
>>   Child process 18746
>> [2013/12/17 16:39:07.845302,  0]
>> winbindd/winbindd_dual.c:1430(fork_domain_child)
>>   adding 0x803358760 to list at 0x803358760
>> [2013/12/17 16:39:07.845302,  0]
>> winbindd/winbindd_dual.c:1430(fork_domain_child)
>>   adding 0x803358760 to list at 0x803358760
>> [2013/12/17 16:39:07.847307,  0] lib/util.c:1117(smb_panic)
>>   PANIC (pid 14996): duplicate!
>>
>> It seems that there are two domains with similar names and winbindd is
>> treating them the same:
>
> Are there really two such domains, or is this an off-by-one
> pointer error somewhere :
>
> EXCHANGE
> -XCHANGE
>  ^
>  |
>  pointer moved forward by one here ?

No, there are two such objects in DNS, but one of them, the EXCHANGE
one, does not respond to us on port 445.

I am getting closer to figuring this out.

-- 
Regards,
Richard Sharpe
(何以解憂？唯有杜康。--曹操)