citgm icon indicating copy to clipboard operation
citgm copied to clipboard

lookup: re-add undici

Open Trott opened this issue 2 years ago • 12 comments

Checklist
  • [x] npm test passes
  • [x] contribution guidelines followed here

Trott avatar Sep 24 '23 22:09 Trott

Codecov Report

All modified lines are covered by tests :white_check_mark:

Comparison is base (f15c9a2) 96.44% compared to head (4211ffe) 96.44%. Report is 12 commits behind head on main.

Additional details and impacted files
@@           Coverage Diff           @@
##             main     #982   +/-   ##
=======================================
  Coverage   96.44%   96.44%           
=======================================
  Files          28       28           
  Lines        2139     2139           
=======================================
  Hits         2063     2063           
  Misses         76       76           

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

codecov-commenter avatar Sep 24 '23 22:09 codecov-commenter

CITGM: https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker/3273/

mcollina avatar Sep 28 '23 11:09 mcollina

Fresh CITGM: https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker/3303/

mcollina avatar Oct 04 '23 07:10 mcollina

Fresh CITGM: https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker/3334/

mcollina avatar Oct 19 '23 08:10 mcollina

Failure from https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker/3334/nodes=rhel8-x64/

 # Subtest: test/connect-timeout.js
     # Subtest: priotorise socket errors over timeouts
         1..1
         not ok 1 - should be equal
           ---
           compare: ===
           at:
             line: 16
             column: 9
             file: test/connect-timeout.js
           stack: |
             test/connect-timeout.js:16:9
           source: |2
                 .catch((err) => {
                   t.equal(err.code, 'ENOTFOUND')
             --------^
                 })
           diff: |
             --- expected
             +++ actual
             @@ -1,1 +1,1 @@
             -ENOTFOUND
             +UND_ERR_CONNECT_TIMEOUT
           ...
         
         # failed 1 test
     not ok 1 - priotorise socket errors over timeouts # time=1074.305ms
     
     # Subtest: connect-timeout
         1..1
         ok 1 - type is ConnectTimeoutError
     ok 2 - connect-timeout # time=1003.865ms
     
     # Subtest: connect-timeout
         1..1
         ok 1 - type is ConnectTimeoutError
     ok 3 - connect-timeout # time=1003.412ms
     
     1..3
     # failed 1 of 3 tests
     # time=5100.208ms
 not ok 30 - test/connect-timeout.js # time=5100.208ms
   ---
   env: {}
   file: test/connect-timeout.js
   timeout: 60000
   command: /home/iojs/build/workspace/citgm-smoker/smoker/bin/node
   args:
     - --expose-gc
     - test/connect-timeout.js
   stdio:
     - 0
     - pipe
     - 2
   cwd: /home/iojs/tmp/citgm_tmp/1031071a-c640-4205-a5f9-5946076938c2/undici
   exitCode: 1
   ...

targos avatar Oct 19 '23 10:10 targos

@nodejs/undici I need some help in making the tests less flaky so we can re-land this.

mcollina avatar Oct 23 '23 07:10 mcollina

I think you fixed connect-timeout? What tests are still flaky?

ronag avatar Oct 23 '23 07:10 ronag

Let's try again with v5.26.5.

Fresh CITGM: https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker/3337/

mcollina avatar Oct 23 '23 07:10 mcollina

Full CITGM for undici https://ci.nodejs.org/job/citgm-smoker/3383/

(I'm not sure if I'm configuring it correctly)

RafaelGSS avatar Jan 16 '24 12:01 RafaelGSS

@RafaelGSS You used the wrong job. It should be https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker-pipeline/

targos avatar Jan 16 '24 12:01 targos

Note this will fail due to the regression in Node v21.6.0

This fixes it but I'm not sure it's the right fix: https://github.com/nodejs/undici/pull/2617

mcollina avatar Jan 16 '24 12:01 mcollina

https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker-pipeline/237/

(v18 and v20)

The undici test suite appears too flaky to be in citgm in its current state.

panva avatar Jan 16 '24 15:01 panva

We need to revisit this. Another regression appeared in v21.7.0 and would have been caught by undici.

targos avatar Mar 07 '24 08:03 targos

IMHO this should never have been removed in the first place (I said as much multiple times). Let's land this even if it fails and refine.

mcollina avatar Mar 07 '24 09:03 mcollina