gentest against s4?

[Volker Lendecke]

Hi, Tridge!

While working on the tevent base infrastructure I noticed
that with new code I've written "make test" stumbles over
gentest.

With my new code it fails with

Notify count inconsistent 0 1
Failed at operation 42 - NTRENAME

If I add a

                if (op == 42) smb_msleep(100);

in line 2939 of gentest.c, it gets much further. Then it
will fail later with a similar message. It seems that this
either strictly depends on timing, or it depends on the
order epoll happens to handle the sockets.

Then I ran gentest without my changes. Make test only
reports success with plain master because it only does 100
operations. If I run gentest similarly to what make test
does, for me it fails consistenly with

Mismatch in internal_information.out.file_id - 0x808000a2ace 0x808000a2acf
Failed at operation 244 - QFILEINFO

So if gentest on the build farm did, say, 245 or more
operations, it would either become a flaky test or plain
fail always. This probably depends on the environment
somehow.

I will not get my code past autobuild due to this.

How are we supposed to handle this situation?

With best regards,

Volker Lendecke

-- 
SerNet GmbH, Bahnhofsallee 1b, 37081 Göttingen
phone: +49-551-370000-0, fax: +49-551-370000-9
AG Göttingen, HRB 2816, GF: Dr. Johannes Loxen