On 2019-03-07 02:47, Aki Tuomi via dovecot wrote:

We'll look into it. It's not usually failing for us, is it failing consistently or randomly?

It is failing consistently. I ran the tests 4 times. It always errors out at the same tests with the same error messages.

Going to bed now. Will be able to answer any other questions you might have in 6-7h.

Cheers, K. C.

-- regards Helmut K. C. Tessarek KeyID 0x172380A011EF4944 Key fingerprint = 8A55 70C1 BD85 D34E ADBC 386C 1723 80A0 11EF 4944

/* Thou shalt not follow the NULL pointer for chaos and madness await thee at its end. */

Thank you very much for getting back to me.

On 2019-03-07 04:23, Stephan Bosch via dovecot wrote:

...

http payload echo (ssl): sequential .................................. : ok http payload echo (ssl): pipeline .................................... : ok http payload echo (ssl): parallel .................................... : FAILED: Test is hanging http payload echo (ssl) .............................................. : FAILED http payload echo (ssl; unknown size): sequential .................... : ok http payload echo (ssl; unknown size): pipeline ...................... : ok http payload echo (ssl; unknown size): parallel ...................... : FAILED: Test is hanging http payload echo (ssl; unknown size) ................................ : FAILED 4 / 114 tests failed Failed to run: ./test-http-payload

The tests ran fine for 2.3.4.1. Is there any way, to debug the issue any further? Do the tests write log files?

It just doesn't make amy sense. Nothing has changed on the OS since I compiled and tested 2.3.4.1.

SSL tests were added here for 2.3.5. Is the test taking a very long time?

No, they don't take very long.

Is this a very slow system?

Define 'very slow`.

$ cat /proc/cpuinfo

[6/6] processor : 0 vendor_id : AuthenticAMD cpu family : 16 model : 6 model name : AMD Turion(tm) II Neo N40L Dual-Core Processor stepping : 3 microcode : 0x10000c8 cpu MHz : 800.000 cache size : 1024 KB physical id : 0 siblings : 2 core id : 0 cpu cores : 2 apicid : 0 initial apicid : 0

                                fpu             : yes

fpu_exception : yes cpuid level : 5 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc rep_good nopl nonstop_tsc extd_apicid pni monitor cx16 popcnt lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt nodeid_msr hw_pstate npt lbrv sv m_lock nrip_save bogomips : 2995.08 TLB size : 1024 4K pages clflush size : 64 cache_alignment : 64 address sizes : 48 bits physical, 48 bits virtual power management: ts ttp tm stc 100mhzsteps hwpstate

processor : 1 vendor_id : AuthenticAMD cpu family : 16 model : 6 model name : AMD Turion(tm) II Neo N40L Dual-Core Processor stepping : 3 microcode : 0x10000c8 cpu MHz : 800.000 cache size : 1024 KB physical id : 0 siblings : 2 core id : 1 cpu cores : 2 apicid : 1 initial apicid : 1

                                fpu             : yes

fpu_exception : yes cpuid level : 5 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc rep_good nopl nonstop_tsc extd_apicid pni monitor cx16 popcnt lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt nodeid_msr hw_pstate npt lbrv sv m_lock nrip_save bogomips : 2995.08 TLB size : 1024 4K pages clflush size : 64 cache_alignment : 64 address sizes : 48 bits physical, 48 bits virtual power management: ts ttp tm stc 100mhzsteps hwpstate

Does this work?:

NOVALGRIND=yes make -C src/lib-http check

Yes, no errors.

Can you make a log of the output of src/lib-http/test-http-payload -D?

Hmm, you could have mentioned that the debug output goes to stderr and not stdout. No errors with that either. If you still need a log, please let me know.

If that doesn't fail for some reason, can you run it within valgrind?

I ran the following command:

valgrind --log-file=valgrind_test-http-payload.out src/lib-http/test-http-payload -D

But I guess the valgrind.log itself is useless to you, and the debug output is printed to the terminal, so I kindliy ask you to be a bit more specific. What exact output do you need or better yet, which command do you want me to run.

Cheers, K. C.

-- regards Helmut K. C. Tessarek KeyID 0x172380A011EF4944 Key fingerprint = 8A55 70C1 BD85 D34E ADBC 386C 1723 80A0 11EF 4944

/* Thou shalt not follow the NULL pointer for chaos and madness await thee at its end. */

Op 08/03/2019 om 23:49 schreef Helmut K. C. Tessarek:

On 2019-03-07 17:03, Helmut K. C. Tessarek via dovecot wrote:

...

...

Does this work?:

NOVALGRIND=yes make -C src/lib-http check Yes, no errors.

Does this mean dovecot will run properly?

Yes. This is test is just a bit unreliable it seems.

Regards,

Stephan.

Op 07/03/2019 om 23:03 schreef Helmut K. C. Tessarek:

Thank you very much for getting back to me.

On 2019-03-07 04:23, Stephan Bosch via dovecot wrote:

...

If that doesn't fail for some reason, can you run it within valgrind? I ran the following command:

valgrind --log-file=valgrind_test-http-payload.out src/lib-http/test-http-payload -D

But I guess the valgrind.log itself is useless to you, and the debug output is printed to the terminal, so I kindliy ask you to be a bit more specific. What exact output do you need or better yet, which command do you want me to run.

Since you're compiling it anyway, maybe you should first try to increase the CLIENT_PROGRESS_TIMEOUT in src/lib-http/test-http-payload.c. It is currently 10 seconds.

Regards,

Stephan.

On Mon, 18 Mar 2019 17:14:24 -0400 "Helmut K. C. Tessarek via dovecot" <dovecot@dovecot.org> wrote:

On 2019-03-08 18:40, Stephan Bosch via dovecot wrote:

...

Since you're compiling it anyway, maybe you should first try to increase the CLIENT_PROGRESS_TIMEOUT in src/lib-http/test-http-payload.c. It is currently 10 seconds.

I've increased the value to 30. This worked. No more timeout.

However, I also tried to link dovecot against a self compiled dynamic openssl 1.1.1 and in that case, there's no issue at all. (With timeout value 10.)

Another issue is that linking against a static openssl 1.1.1 fails miserably. But apparently there are more and more products that ignore the premise that linking against a static openssl is safer. dovecot is not the only SW that does not work with a static openssl variant.

Just a guess ... maybe it is lack of entropy on the build machine? Try running haveged.

-- openSUSE - SUSE Linux is my linux openSUSE is good for you www.opensuse.org