[MDEV-16075] CTests failing Created: 2018-05-02  Updated: 2018-07-02  Resolved: 2018-07-02

Status: Closed
Project: MariaDB Server
Component/s: Tests
Affects Version/s: 10.2.0
Fix Version/s: 5.5.62

Type: Bug Priority: Major
Reporter: Teodor Mircea Ionita (Inactive) Assignee: Teodor Mircea Ionita (Inactive)
Resolution: Won't Fix Votes: 0
Labels: None
Environment:

Ubuntu 14.04+, macOS 10.13


Issue Links:
Duplicate
duplicates MDEV-12470 'make test' failing Closed

 Description   

It looks like a couple of the CTests that are ran with make test constantly fail on macOS, Linux for the 10.0 branch:

      Start 23: bulk1
23/80 Test #23: bulk1 ............................***Failed    0.07 sec
      Start 24: performance
24/80 Test #24: performance ......................***Failed    0.06 sec
      Start 25: basic-t
25/80 Test #25: basic-t ..........................***Failed    0.07 sec
      Start 26: fetch
26/80 Test #26: fetch ............................***Failed    0.07 sec
      Start 27: charset
27/80 Test #27: charset ..........................***Failed    0.06 sec
      Start 28: logs
28/80 Test #28: logs .............................***Failed    0.06 sec
      Start 29: cursor
29/80 Test #29: cursor ...........................***Failed    0.07 sec
      Start 30: errors
30/80 Test #30: errors ...........................***Failed    0.06 sec
      Start 31: view
31/80 Test #31: view .............................***Failed    0.07 sec
      Start 32: ps
32/80 Test #32: ps ...............................***Failed    0.07 sec
      Start 33: ps_bugs
33/80 Test #33: ps_bugs ..........................***Failed    0.07 sec
      Start 34: sp
34/80 Test #34: sp ...............................***Failed    0.06 sec
      Start 35: result
35/80 Test #35: result ...........................***Failed    0.07 sec
      Start 36: connection
36/80 Test #36: connection .......................***Failed    0.06 sec
      Start 37: misc
37/80 Test #37: misc .............................***Failed    0.07 sec
      Start 38: ps_new
38/80 Test #38: ps_new ...........................***Failed    0.07 sec
      Start 39: thread
39/80 Test #39: thread ...........................***Failed    0.06 sec
      Start 40: features-10_2
40/80 Test #40: features-10_2 ....................***Failed    0.08 sec
      Start 41: dyncol
41/80 Test #41: dyncol ...........................***Failed    0.06 sec
      Start 42: async
42/80 Test #42: async ............................***Failed    0.06 sec

Is this a known issue? How is running make test different than ./mtr --suite=unit? Is running make test advised in any case in favor of mtr?



 Comments   
Comment by Teodor Mircea Ionita (Inactive) [ 2018-05-02 ]

It looks like the failing tests were introduced in 10.2.

Comment by Sergei Golubchik [ 2018-05-02 ]

./mtr --suite=unit starts mysqld that unit tests are using. When you run them standalone, they fail as they cannot find the server.

Comment by Teodor Mircea Ionita (Inactive) [ 2018-05-02 ]

In that case, wouldn't it be better if make test would run MTR in the first place? Or at least print a useful message?

Comment by Sergei Golubchik [ 2018-05-02 ]

Absolutely. That would be ideal.

Comment by Vicențiu Ciorbaru [ 2018-05-03 ]

We should properly fix this, just to make life easier for certain distributions.

There is a problem with overriding `make test` command in cmake. CMake will issue a warning when this is done by creating a new target. There should be a "proper" way to get it working.

Comment by Teodor Mircea Ionita (Inactive) [ 2018-05-07 ]

Managed to come up with a solution after quite a bit of headache. See PR at https://github.com/MariaDB/server/pull/746.

It is based on the solution explained here https://stackoverflow.com/questions/31252228/cmake-and-ctests-default-test-command-skips-a-specially-named-test

Comment by Teodor Mircea Ionita (Inactive) [ 2018-05-17 ]

serg cvicentiu bump ^

Comment by Sergei Golubchik [ 2018-05-17 ]

Yes, I've seen it. Very nice trick. Looks like the best we could do...

Comment by Teodor Mircea Ionita (Inactive) [ 2018-07-02 ]

Closing this since, as discussed, the fix in PR#746 was merged and reverted due to insufficient verbosity while running for a long time and no other solution is likely possible at the present moment with cmake/ctest 3.11.

Generated at Thu Feb 08 08:26:11 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.