Re: kern/53161: ATF test runs leave a rump_server process around

[Ryota Ozaki <ozaki-r%netbsd.org@localhost>]

On Fri, Apr 6, 2018 at 5:14 PM, Ryota Ozaki <ozaki-r%netbsd.org@localhost> wrote:
> On Fri, Apr 6, 2018 at 5:00 PM,  <martin%netbsd.org@localhost> wrote:
>>>Number:         53161
>>>Category:       kern
>>>Synopsis:       ATF test runs leave a rump_server process around
>>>Confidential:   no
>>>Severity:       critical
>>>Priority:       high
>>>Responsible:    kern-bug-people
>>>State:          open
>>>Class:          sw-bug
>>>Submitter-Id:   net
>>>Arrival-Date:   Fri Apr 06 08:00:00 +0000 2018
>>>Originator:     Martin Husemann
>>>Release:        NetBSD 8.0_BETA
>>>Organization:
>> The NetBSD Foundation, Inc.
>>>Environment:
>> System: NetBSD unpluged.duskware.de 8.0_BETA NetBSD 8.0_BETA (UNPLUGED) #30: Thu Apr 5 17:47:24 CEST 2018 martin%seven-days-to-the-wolves.aprisoft.de@localhost:/work/src-8/sys/arch/evbarm/compile/UNPLUGED evbarm
>> Architecture: earm
>> Machine: evbarm
>>>Description:
>>
>> Since one or two weeks, on evbarm, after a full ATF test run, a rump_server
>> process is left running, eating lots of CPU.
>>
>> I have not identified the test triggering this yet. This does not happen in
>> -current.
>
> The last argument of the rump_server process may tell us which test
> starts the rump_server. It's a unix domain socket for rump kernels
> and many tests embed its test name into the filename.

Oops. This is true only for tests/net. If the rump_server is started from
tests of other categories, I have no idea to identify.

  ozaki-r