You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@airflow.apache.org by GitBox <gi...@apache.org> on 2021/03/21 09:25:06 UTC

[GitHub] [airflow] potiuk edited a comment on pull request #14915: Running tests in parallel

potiuk edited a comment on pull request #14915:
URL: https://github.com/apache/airflow/pull/14915#issuecomment-803540351


   Monitoring helped a lot:
   
   * I already decreased the use of MySQL memory from ~ 450 MB in peak to ~200MB. The default MySQL configuration we used was not optimised at all for our test usage
   * However it seems that we have a serious problem in our CLI Tests that often spins out of control:
   
   https://github.com/potiuk/airflow/runs/2158887308?check_suite_focus=true#step:6:4589
   
   Usually our test `airflow` container max at around 400MB, but in cases where we have Exit 137 and failed tests, those spin out of control a lot a
   
   ```
     Overall resource statistics
     
     CONTAINER ID                                                       NAME                                         CPU %     MEM USAGE / LIMIT     MEM %     NET I/O           BLOCK I/O        PIDS
     7a4dffd2e4123a8f567704e22d13a41010c3e3f56f6759eac0c56dbb02138bea   airflow-providers_airflow_run_92452ab2adba   22.43%    643MiB / 6.791GiB     9.25%     2.21MB / 2.66MB   21.7MB / 307kB   8
     7605d1260aea0e05c3b5c927a43f85bfd0c89f9165e59218d9b7b5773caf84c1   airflow-providers_mysql_1                    0.20%     94.01MiB / 6.791GiB   1.35%     2.66MB / 2.21MB   36.9kB / 432MB   32
     88d5fd7c47ef1dc1c1269a488b2800183c262a68a3b73a6333bebff2b365e288   airflow-cli_airflow_run_17f7dab0ae27         142.86%   1.829GiB / 6.791GiB   26.93%    9.15MB / 8.32MB   15.1MB / 479kB   54
     c1141763d37ed964f6d6f76e6f842c45e0bc99d5d91d323b3bd069b456c6511d   airflow-cli_mysql_1                          0.19%     104.8MiB / 6.791GiB   1.51%     8.32MB / 9.14MB   684kB / 467MB    38
   ````
   
   See 1.829GB use of CLI airflow container
   
   and then 
   
   ```
     CONTAINER ID                                                       NAME                                         CPU %     MEM USAGE / LIMIT     MEM %     NET I/O           BLOCK I/O        PIDS
     7a4dffd2e4123a8f567704e22d13a41010c3e3f56f6759eac0c56dbb02138bea   airflow-providers_airflow_run_92452ab2adba   58.75%    633.5MiB / 6.791GiB   9.11%     2.26MB / 2.71MB   31.6MB / 307kB   8
     7605d1260aea0e05c3b5c927a43f85bfd0c89f9165e59218d9b7b5773caf84c1   airflow-providers_mysql_1                    4.69%     90MiB / 6.791GiB      1.29%     2.71MB / 2.26MB   1.26MB / 432MB   32
     88d5fd7c47ef1dc1c1269a488b2800183c262a68a3b73a6333bebff2b365e288   airflow-cli_airflow_run_17f7dab0ae27         71.13%    3.781GiB / 6.791GiB   55.68%    9.32MB / 8.43MB   16.1MB / 483kB   42
     c1141763d37ed964f6d6f76e6f842c45e0bc99d5d91d323b3bd069b456c6511d   airflow-cli_mysql_1                          0.05%     101.4MiB / 6.791GiB   1.46%     8.43MB / 9.32MB   684kB / 469MB    38
   ```
   
   Then it gets to 3.78 GB
   
   This is always around this state of CLI tests:
   
   ```
     ### The last lines for CLI process ###
     
     tests/cli/commands/test_sync_perm_command.py .                           [ 67%]
     tests/cli/commands/test_task_command.py ..........s...........
   ```
   
   so I think it will be rather easy to find the culprit.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org