s4:torture: Increase multichannel timeout

[Joseph Sutton]

On 19/10/23 2:54 pm, Joseph Sutton via samba-technical wrote:
> On 18/10/23 7:28 pm, Andreas Schneider via samba-technical wrote:
>> On Tuesday, 17 October 2023 22:23:02 CEST Andrew Bartlett wrote:
>>> The branch, master has been updated
>>>         via  f392fdfd47f shadow_copy2: Add missing TALLOC_FREE
>>>         via  c6d0df787a1 s4:torture: Increase multichannel timeout
>>
>> Hi,
>>
>> I've increased the timeout which worked for me and the test didn't run 
>> into
>> timeouts anymore. However Joseph run into the issue again with
>>
>> https://gitlab.com/samba-team/devel/samba/-/jobs/5316200327
>>
>> Does it make sense to further increase the timeout or could there be 
>> another
>> issue. I'm not an expert in this area of the code.
>>
>> https://gitlab.com/samba-team/samba/-/commit/
>> c6d0df787a1f6007e1f4594f68ff1f75a46bd293
>>
> 
> Here’s some more detailed output from a failed run:
> https://gitlab.com/samba-team/devel/samba/-/jobs/5324774640
> 
> Testing for BUG 15346
> 2023-10-19T01:09:58.811178Z: conn[0]: negprot done
> 2023-10-19T01:09:58.811246Z: conn[3]: negprot done
> 2023-10-19T01:09:58.811290Z: conn[5]: negprot done
> […]
> 2023-10-19T01:09:58.883395Z: conn[30]: echo done
> 2023-10-19T01:11:31.861815Z: conn[3]: echo timed out
> UNEXPECTED(failure): samba3.smb2.multichannel.bugs.bug_15346(nt4_dc)
> REASON: Exception: Exception: 
> ../../source4/torture/smb2/multichannel.c:2474: status was 
> NT_STATUS_IO_TIMEOUT, expected NT_STATUS_OK: smb2cli_echo_recv failed
> 
> Connection 3 took just over 93 seconds to time out. 93 happens to be 
> equal to ‘state->num_conns * 3’, the current timeout value in seconds.
> 
> So it would appear that the failure is indeed due to this timeout. But 
> when the timeout is so long already, I don’t think that increasing it 
> further is going to help.
> 

Also, when the test fails, it is nearly always the second connection to 
complete protocol negotiation that gets stuck and eventually times out. 
But in [0], it is the twelfth (connection 12). And in [1], both the 
second (connection 1) and the fourth (connection 4) appear to have timed 
out.

Regards,
Joseph

[0] https://gitlab.com/samba-team/devel/samba/-/jobs/5300258543

[1] https://gitlab.com/samba-team/devel/samba/-/jobs/5301618462