Closed
Bug 1200470
Opened 10 years ago
Closed 10 years ago
Please deploy autopush 1.4.1 to PRODUCTION
Categories
(Cloud Services :: Operations: Deployment Requests - DEPRECATED, task)
Cloud Services
Operations: Deployment Requests - DEPRECATED
Tracking
(Not tracked)
VERIFIED
FIXED
People
(Reporter: lina, Assigned: oremj)
References
Details
Including all changes since 1.3.0. Richard, Jeremy—could we skip 1.4.0 and deploy 1.4.1 directly, please?
1.4.1 (2015-08-31)
==================
Bug Fixes
---------
* Expose Web Push headers for CORS requests. PR #148.
* Expose argument for larger websocket message sizes (to fix issue #151)
Clients with a large number of channelIDs (50+) can cause the initial
connection to fail. A proper solution is to modify the client to not send
ChannelIDs as part of the "hello" message, but being able to increase the
message size on the server should keep the server from dying up front.
This fix should only impact clients with large numbers of registered channels,
notably, devs.
1.4.0 (2015-08-27)
==================
Bug Fixes
---------
* Fix _notify_node to not attempt delivering to ourselves at the end of the
client connection.
* Remove adaptive ping entirely. Send special close code and drop clients that
ping more frequently than 55 seconds (approx 1 min). This will result in
clients that ping too much being turned away for awhile, but will alleviate
data/battery issues in buggy mobile clients. Issue #103.
* Store and transmit encrypted Web Push messages as Base64-encoded strings.
PR #135.
Features
--------
* Add /status HTTP endpoint for autopush. Issue #136.
* Log all disconnects, whether they were clean, the code, and the reason.
* Allow encryption headers to be omitted for blank messages. Issue #132.
1.3.3 (2015-08-18)
==================
* Handle None values in ack updates.
1.3.2 (2015-08-11)
==================
Bug Fixes
---------
* Fix deferToLater to not call the function if it was cancelled using a
canceller function.
* Fix finish_webpush_notifications to not immediately call
process_notifications as that will be called as needed after ack's have been
completed.
* Fix process_ack to not call process_notifications when using webpush if there
are still remaining notifications to ack.
Features
--------
* Integrate simplepush_test smoke-test client with the main autopush test-suite
into the test-runner. Issue #119.
1.3.1 (2015-08-04)
==================
Bug Fixes
---------
* Fix RouterException to allow for non-logged responses. Change
RouterException's to only log actual exceptions that should be address in
bug-fixes. Issue #125.
1.3.0 (2015-07-29)
==================
Features
--------
* Add WebPush TTL scheme per spec (as of July 28th 2015). Issue #56.
* Add WebPush style data delivery with crypto headers to connected clients.
Each message is stored independently in a new message table, with the version
and channel id still required to ack a message. The version is a UUID4 hex
which is also echo'd back to the AppServer as a Location URL per the current
WebPush spec (as of July 28th 2015). Issue #57.
* Add Sphinx docs with ReadTheDocs publishing. Issue #98.
This change also includes a slight Metrics refactoring with a IMetrics
interface, and renames MetricSink -> SinkMetrics for naming consistency.
Bug Fixes
---------
* Increase test coverage of utils for 100% test coverage.
* Move all dependencies into requirements.txt and freeze them all explicitly.
Backward Incompatibilities
--------------------------
* ``bridge`` option is now ``external_router``.
Comment 2•10 years ago
|
||
Hey Kit, I'm OK with skipping 0.4.0 and moving straight to 0.4.1, but shouldn't these changes be staged and tested first? I can do a quick turn-around if :oremj has time to stage these this morning.
Flags: needinfo?(kcambridge)
| Reporter | ||
Comment 3•10 years ago
|
||
Oops, forgot to block on bug 1200468; sorry about that. There's a config change in this release, so yes, we should definitely stage first.
Flags: needinfo?(kcambridge)
Comment 4•10 years ago
|
||
pre-deployment...
----------------------------
URL CHECKS (PRE-PROD)
----------------------------
https://push.services.mozilla.com:
Response time: 0.122708
https://updates.push.services.mozilla.com/health:
{
"status": "OK",
"prod.autopush.storage": {
"status": "OK"
},
"version": "1.2.3",
"clients": 0,
"prod.autopush.router": {
"status": "OK"
}
}
Response time: 0.130014
https://updates.push.services.mozilla.com/status:
{
"status": "OK",
"version": "1.2.3"
}
Response time: 0.126463
Comment 5•10 years ago
|
||
----------------------------
SUMMARY
----------------------------
oremj has deployed this release to production and DNS has been switched.
E2E TESTS
Able to successfully locate, ring, lock and erase Flame device (w/ B2G 2.2) using findmydevice application
x10
Able to place several Hello (Loop) calls successfully between Nightly (43.0a1) and GR (40.0.3) w/ autopush
and verify e2e test calls using FxA contacts, room notifications - OK
Per :benbangert, everything working
Verifying this release.
----------------------------
URL CHECKS (PROD)
----------------------------
https://push.services.mozilla.com:
Response time: 0.116556
https://updates.push.services.mozilla.com/health:
{
"status": "OK",
"prod.autopush.storage": {
"status": "OK"
},
"version": "1.4.1",
"clients": 0,
"prod.autopush.router": {
"status": "OK"
}
}
Response time: 0.126496
https://updates.push.services.mozilla.com/status:
{
"status": "OK",
"version": "1.4.1"
}
Status: NEW → RESOLVED
Closed: 10 years ago
Resolution: --- → FIXED
Updated•10 years ago
|
Status: RESOLVED → VERIFIED
You need to log in
before you can comment on or make changes to this bug.
Description
•