@jskladan complained about information overload in our "daily" (twice-daily) taskotron status emails, and he's right. We need to make sure that only interesting and important information are displayed, otherwise people quickly start to ignore these status reports. Here are some ideas of what we would like to change:
limit
1000
Example subject:
Taskotron status: DEV from 2014-08-26
Example email body:
Executed checks --------------- depcheck: 579 (11% failed) upgradepath: 184 (17% failed) rpmlint: 2158 (34% failed) Failure links ------------- depcheck: http://taskotron-dev.fedoraproject.org/resultsdb/results?outcome=FAILED,ERROR,WAIVED,NEEDS_INSPECTION&since=2014-08-26T00:00:00,2014-08-27T00:00:00&testcase_name=depcheck&limit=64 upgradepath: http://taskotron-dev.fedoraproject.org/resultsdb/results?outcome=FAILED,ERROR,WAIVED,NEEDS_INSPECTION&since=2014-08-26T00:00:00,2014-08-27T00:00:00&testcase_name=upgradepath&limit=32 rpmlint: http://taskotron-dev.fedoraproject.org/resultsdb/results?outcome=FAILED,ERROR,WAIVED,NEEDS_INSPECTION&since=2014-08-26T00:00:00,2014-08-27T00:00:00&testcase_name=rpmlint&limit=734
Note: We don't currently use ERROR and other status codes. Once we do, it will be trivial to append this information to the email body, and provide also error links, just not failure links.
ERROR
@jsedlak, can you please implement necessary changes in fetch_activity.py?
fetch_activity.py
Since ansible-playbooks aren't configured in phabricator, I have pushed my changes into bitbucket into new branch [[ https://bitbucket.org/fedoraqa/ansible-playbooks/commits/2494aadf7cc974c23cbf31ca46d2cc55039ddae4 | here ]]. How often are emails sent and what's in the subject is trigger's responsibility.
Send the emails once per day, not twice per day. The status report should always cover the previous day, midnight to midnight. Set a reasonable email subject, because it's hard to see from the current cryptic subject what it is about.
Just finished the above and deployed to dev and stg.
@jsedlak: Please round the number after division. Can you provide a sample output?
Also, please see the built-in min function, to be used here:
min
+ limit = 1000 if failed_types[failed_check] > 1000 else failed_types[failed_check]
I have [[ https://bitbucket.org/fedoraqa/ansible-playbooks/commits/ffe2a1c256b565699b98d9e97426b4330433fddf | updated script ]] on bitbucket.
Here is output for sample data from resultsdb's init_db.sh script.
init_db.sh
libtaskotron results ==================== Generated on: dhcp-29-204.brq.redhat.com From: 2014-09-07T22:59:24.420151 To: 2014-09-08T10:59:24.420151 Executed checks: ---------------- depcheck: 2 (50 % failed) rpmlint: 1 (100 % failed) Total: 3 executed, 2 failed Links to failed checks: ----------------------- depcheck: http://localhost:5001/results?outcome=FAILED,ERROR,WAIVED,NEEDS_INSPECTION&since=2014-09-07T22:59:24.420151,2014-09-08T10:59:24.420151&testcase_name=depcheck&limit=1 rpmlint: http://localhost:5001/results?outcome=FAILED,ERROR,WAIVED,NEEDS_INSPECTION&since=2014-09-07T22:59:24.420151,2014-09-08T10:59:24.420151&testcase_name=rpmlint&limit=1
Looks good.
@mkrizek, can we please deploy this?
Deployed to dev.
Works on dev, deployed to stg. Closing.
Metadata Update from @kparal: - Issue tagged with: infrastructure
Log in to comment on this ticket.