Details
Major
Description
With 10.0.16 the unit tests ma_test_loghandler and ma_test_loghandler_readonly are failing..
31: ok 7987 - write LOGREC_FIXED_RECORD_2LSN_EXAMPLE 31: ok 7988 - write LOGREC_VARIABLE_RECORD_2LSN_EXAMPLE 31: ok 7989 - write LOGREC_FIXED_RECORD_0LSN_EXAMPLE 31: ok 7990 - write LOGREC_VARIABLE_RECORD_0LSN_EXAMPLE 31: ok 7991 - flush 31: ok 7992 - write LOGREC_FIXED_RECORD_1LSN_EXAMPLE 31: ok 7993 - write LOGREC_VARIABLE_RECORD_1LSN_EXAMPLE 31: ok 7994 - write LOGREC_FIXED_RECORD_0LSN_EXAMPLE 31: ok 7995 - write LOGREC_VARIABLE_RECORD_0LSN_EXAMPLE 31: ok 7996 - flush 31: ok 7997 - flush 31: ok 7998 - read record type: 1 LSN: (1,0x201b) 31: ok 7999 - read record type: 3 LSN: (1,0x2024) 31: Incorrect LOGREC_VARIABLE_RECORD_1LSN_EXAMPLE data read(1)type 4 (0), strid 1 (0), len 24582, 39594 + 7 (1), hdr len: 12 (0), ref(1,0x201b), lsn(1,0x2029) (0), content: 0 31: not ok 8000 - read record 31: # 14394 tests planned but only 8000 executed 31/55 Test #31: ma_test_loghandler ...............***Failed 3.29 sec
40: 1..10 40: ok 1 - write LOGREC_FIXED_RECORD_0LSN_EXAMPLE 40: ok 2 - write LOGREC_FIXED_RECORD_1LSN_EXAMPLE 40: ok 3 - write LOGREC_VARIABLE_RECORD_1LSN_EXAMPLE 40: ok 4 - write LOGREC_FIXED_RECORD_0LSN_EXAMPLE 40: ok 5 - write LOGREC_VARIABLE_RECORD_0LSN_EXAMPLE 40: ok 6 - read record 40: ok 7 - read record 40: not ok 8 - read record 40: Incorrect LOGREC_VARIABLE_RECORD_1LSN_EXAMPLE data read(1)type 4 (0), strid 1 (0), len 318041, 380658 + 7 (1), hdr len: 12 (0), ref(1,0x2007), lsn(1,0x2015) (0), content: 0 40: # 10 tests planned but only 8 executed 40/55 Test #40: ma_test_loghandler_readonly ......***Failed 0.05 sec
Gliffy Diagrams
Attachments
Activity
- All
- Comments
- Work Log
- History
- Activity
- Transitions
With .14 and back through to .7 there have been no unit test failures. I don't remember if I ran the test suite on .15. So having any failures within the last few releases is new.
Yes, with the means of building any packages built and the test is run from a clean build.
I am seeing these failures on amd64 powerpc and sparc64 so they do not seem to be arch specific.
I see, thanks. What about your build options and how you run the tests (via make test, or manually, or...)?
It's for me to try to reproduce the problem locally.
Closing for now as incomplete, please comment to re-open if you have more information.
The closest I have handy for a relatively quick verification is FreeBSD, it doesn't seem to fail for me there.
Is it reproducible on a clean build?
Is it 386 or x86_64? (I have no idea if it's important, but it might be)
Could you please describe how exactly you build MariaDB (all cmake options) and how you run the tests? Maybe there's something in it.