Inspect flaky test-net-bytes-per-incoming-chunk-overhead #23740

Activity

1. Trott

   added

   netIssues and PRs related to the net subsystem.Issues and PRs related to the net subsystem.

   flaky-testIssues and PRs involving tests that fail intermittently in CI.Issues and PRs involving tests that fail intermittently in CI.

   on Oct 18, 2018
2. Trott commented on Oct 21, 2018

   

   Trott

   on Oct 21, 2018

   MemberAuthor

   More actions

   https://ci.nodejs.org/job/node-test-commit-linux/22513/nodes=ubuntu1804-docker/console

   19:34:03 not ok 2375 sequential/test-net-bytes-per-incoming-chunk-overhead
   19:34:03   ---
   19:34:03   duration_ms: 120.106
   19:34:03   severity: fail
   19:34:03   exitcode: -15
   19:34:03   stack: |-
   19:34:03     timeout
   19:34:03   ...

   Looks like it's a particular problem in the docker container. Maybe @nodejs/docker would have some insight.
3. Trott commented on Oct 21, 2018

   

   Trott

   on Oct 21, 2018

   MemberAuthor

   More actions

   It does seem like something is up that slows the docker host down during these timeouts.

   Example run with failure: https://ci.nodejs.org/job/node-test-commit-linux/22513/nodes=ubuntu1804-docker/console

   sequential/test-net-bytes-per-incoming-chunk-overhead times out after 120 seconds
   sequential/test-net-GH-5504 4.806 seconds
   sequential/test-net-listen-shared-ports 4.474 seconds

   Example run with success: https://ci.nodejs.org/job/node-test-commit-linux/22514/nodes=ubuntu1804-docker/console

   sequential/test-net-bytes-per-incoming-chunk-overhead 6.522 seconds
   sequential/test-net-GH-5504 0.510 seconds
   sequential/test-net-listen-shared-ports 0.509 seconds

   Both were on test-digitalocean-ubuntu1804_container-x64-1 but the success was on test-digitalocean-ubuntu1804_container-x64-2 so maybe that's significant?

   Why yes, it probably is. Here's a successful run on test-digitalocean-ubuntu1804_container-x64-1:

   https://ci.nodejs.org/job/node-test-commit-linux/22516/nodes=ubuntu1804-docker/console:

   sequential/test-net-bytes-per-incoming-chunk-overhead 89.54 seconds
   sequential/test-net-GH-5504 2.655 seconds
   sequential/test-net-listen-shared-ports 2.926 seconds

   It seems to be sustained across multiple builds on test-digitalocean-ubuntu1804_container-x64-1. Here's https://ci.nodejs.org/job/node-test-commit-linux/22518/nodes=ubuntu1804-docker/console:

   sequential/test-net-bytes-per-incoming-chunk-overhead 49.683 seconds

   sequential/test-net-bytes-per-incoming-chunk-overhead:

   sequential/test-net-bytes-per-incoming-chunk-overhead 97.632 seconds

   /ping @nodejs/build is test-digitalocean-ubuntu1804_container-x64-1 provisioned less generously than test-digitalocean-ubuntu1804_container-x64-2? Or maybe it has some intensive stale processes running on it that need to be terminated or something?
4. Trott commented on Nov 23, 2018

   

   Trott

   on Nov 23, 2018

   MemberAuthor

   More actions

   Seems to not be happening anymore. Closing, can re-open if it surfaces again.
5. Trott

   closed this as completedon Nov 23, 2018