Closed Bug 1237669 Opened 10 years ago Closed 10 years ago

Please re-deploy autopush 1.8.1 to STAGE w/ new ELB architecture

Categories

(Cloud Services :: Operations: Deployment Requests - DEPRECATED, task)

task
Not set
normal

Tracking

(Not tracked)

VERIFIED FIXED

People

(Reporter: oremj, Assigned: oremj)

References

Details

No description provided.
:oremj has re-deployed autopush 1.8.1 to STAGE using new ELB backed architecture. we will be deploying this to PROD next Wed., 1/13 @9am PST STAGE test results to follow... ------------------ RELEASE NOTES (1.8.1) ------------------ https://github.com/mozilla-services/autopush/releases COMPARISONS https://github.com/mozilla-services/autopush/compare/1.7.1...1.7.2 https://github.com/mozilla-services/autopush/compare/1.7.2...1.8.0 https://github.com/mozilla-services/autopush/compare/1.8.0...1.8.1 TAGS https://github.com/mozilla-services/autopush/releases/tag/1.8.1 https://github.com/mozilla-services/autopush/commit/7025e1d1293e85f35dedb2bcba214fc3bb881c3b CHANGELOG 1.8.1 (2015-11-16) ================== Features -------- * Convert proprietary AUTH to use Bearer Token for client REST interfaces. Issue #238. Bug Fixes --------- WebPush ------- Configuration Changes --------------------- * Please include the new `--auth_key` which is the base token set for generating bearer tokens. This uses the same format as the `--crypto_key`, but should be a different value to prevent possible key detection. The key can be generated using the same `bin/autokey` tool used to generate the crypto_key
Status: NEW → RESOLVED
Closed: 10 years ago
QA Contact: rpappalardo
Resolution: --- → FIXED
Summary: autopush: move to ELB architecture → Please re-deploy autopush 1.8.1 to STAGE w/ new ELB architecture
---------------------------- HOST CHECKS (STAGE) ---------------------------- EC2 INSTANCE PROPERTIES region: us-east-1: id: i-f312610a tags["Type"]: autopush tags["Stack"]: autopush-dev public_dns_name: ec2-52-91-136-20.compute-1.amazonaws.com launch_time: 2016-01-06T18:14:51.000Z ---------------------------- URL CHECKS (STAGE) ---------------------------- https://updates-autopush.stage.mozaws.net/health: { "status": "OK", "stage.autopush.storage": { "status": "OK" }, "stage.autopush.router": { "status": "OK" }, "version": "1.8.1", "clients": 0 } Response time: 0.411325 https://updates-autopush.stage.mozaws.net/status: { "status": "OK", "version": "1.8.1" } Response time: 0.36469
---------------------------- LOADTESTS ---------------------------- loadtesting OK per :benbangert. notes from IRC. Ben, can you confirm I have this right or add anything I may have missed about your loadtest results? - some memory improved with ELB deployment - average: 1 million connections / 15 boxes @64k / connection node - new ELB backed cluster able to reach 9M Connections Established
Flags: needinfo?(bbangert)
---------------------------- SUMMARY ---------------------------- Everything looks good. Once Ben confirms loadtest results (Comment #3), I'll verify and create PROD ticket. unittests passed - OK ---------------------------- E2E TESTS ---------------------------- DESKTOP PUSH NOTIFICATIONS using test page: https://pdehaan.github.io/push-notification-test/ - OK FINDMYDEVICE Able to successfully locate, ring and lock STAGE-configured Flame device (w/ B2G 2.2) x10 HELLO Able to place several Hello (Loop) calls successfully between Nightly (46.0a1) and GR (44.0b6) w/ autopush and verify e2e test calls, room notifications - OK
(In reply to Richard Pappalardo [:rpapa][:rpappalardo] from comment #3) > ---------------------------- > LOADTESTS > ---------------------------- > > loadtesting OK per :benbangert. notes from IRC. > Ben, can you confirm I have this right or add anything I may have missed > about your loadtest results? > > - some memory improved with ELB deployment > - average: 1 million connections / 15 boxes @64k / connection node > - new ELB backed cluster able to reach 9M Connections Established one correction from ben: 16 boxes not 15 also testing was done using more connections using less nodes (than currently on production) oremj: can you confirm how many nodes will be deployed on backend in production?
Flags: needinfo?(bbangert) → needinfo?(oremj)
We won't be alterning the node count in this iteration. It will still be 30.
Flags: needinfo?(oremj)
Thanks, Jeremy. Verifying STAGE
Status: RESOLVED → VERIFIED
You need to log in before you can comment on or make changes to this bug.