Taskcluster/Meetings/2014-01-23: Difference between revisions

Latest revision as of 18:38, 3 July 2017

Previous Action Items

[dustin] to discuss deployment strategies (heroku vs internal?) with lightsofapollo and other module owners
- Done - will continue to revisit as development proceeds
[jgriffin] to find a different meeting time for next week
- Done, let me know if Thursdays @ 11am is bad for someone.

Status Updates

Job Queue (owner: dminor/ted)

Created postgres based branch
- sqlite was slow even for testing!
- task queue in postgres doesn't work very well and was replaced in socorro with rabbitmq
Created postgres/rabbitmq branch
- Moved job queue to rabbit, tests are working well with minimal changes
- Set up docker packaging to handle postgres, rabbit dependencies
- Working on updating the stresstest to work with amqp
- Looking to merge soon

Provisioner (owner: jonas)

Works:

Creates spot requests if there is pending tasks in queue
Cancels spot requests if pending tasks disappear
Kills instances if more than max-instances is running
Persona login for debug web-interface
Config files with queue hostname, etc.
A few superficial tests

Todo:

Finish up web-interface to see provisioner state, and reasoning behind decisions,
Test in an actual deployment with AMIs that has workers, and the real queue (not the mock queue, I've used so far)
Two-legged oauth at some point...
When we're submitting test tasks to a real system, we can work on improving the provisioning algorithm

Worker (owner: lightsofapollo, jonas, ted)

 - live logging done (as of last week)
 - wiring up existing docker running capabilities to amqp/http interfaces (with live logging either on or off) in progress
 - Aim to have complete (AMI) deploy-able image by next meeting.

Task Graph (owner: aki)

not started, b2g and vcs-sync work. Hopefully will be able to delegate some of that to free up time.

Treeherder Integration (owner: jeads)

End-to-End Testing (ted)

 - initial dev box is ready https://github.com/taskcluster/packer-boxes (and available on s3) usable for docker "service" based testing.
 - need final agreement on docker-ifying the rest of the services and plan for the "deployment" repo
 - timeframe for integration testing?

Round Table

testing of services? (just looking for formal agreement on docker all top level services)
documentation site
re-spawn after try jobs? --dustin

@@ Line 1: / Line 1: @@
 == Previous Action Items ==
+* [dustin] to discuss deployment strategies (heroku vs internal?) with lightsofapollo and other module owners
+** Done - will continue to revisit as development proceeds
+* [jgriffin] to find a different meeting time for next week
+** Done, let me know if Thursdays @ 11am is bad for someone.
 == Status Updates ==
 === Job Queue (owner: dminor/ted) ===
+* Created postgres based branch
+** sqlite was slow even for testing!
+** task queue in postgres doesn't work very well and was replaced in socorro with rabbitmq
+* Created postgres/rabbitmq branch
+** Moved job queue to rabbit, tests are working well with minimal changes
+** Set up docker packaging to handle postgres, rabbit dependencies
+** Working on updating the stresstest to work with amqp
+** Looking to merge soon
 === Provisioner (owner: jonas) ===
+'''Works:'''
+* Creates spot requests if there is pending tasks in queue
+* Cancels spot requests if pending tasks disappear
+* Kills instances if more than <code>max-instances</code> is running
+* Persona login for debug web-interface
+* Config files with queue hostname, etc.
+* A few superficial tests
+'''Todo:'''
+* Finish up web-interface to see provisioner state, and reasoning behind decisions,
+* Test in an actual deployment with AMIs that has workers, and the real queue (not the mock queue, I've used so far)
+* Two-legged oauth at some point...
+* When we're submitting test tasks to a real system, we can work on improving the provisioning algorithm
-=== Worker (owner: lightsofapollo) ===
+=== Worker (owner: lightsofapollo, jonas, ted) ===
+  - live logging done (as of last week)
+  - wiring up existing docker running capabilities to amqp/http interfaces (with live logging either on or off) in progress
+  - Aim to have complete (AMI) deploy-able image by next meeting.
 === Task Graph (owner: aki) ===
+* not started, b2g and vcs-sync work.  Hopefully will be able to delegate some of that to free up time.
 === Treeherder Integration (owner: jeads) ===
+=== End-to-End Testing (ted) ===
+  - initial dev box is ready https://github.com/taskcluster/packer-boxes (and available on s3) usable for docker "service" based testing.
+  - need final agreement on docker-ifying the rest of the services and plan for the "deployment" repo
+  - timeframe for integration testing?
 == Round Table ==
+* testing of services? (just looking for formal agreement on docker all top level services)
+* documentation site
+* re-spawn after try jobs? --dustin
 == Action Items ==