aurora-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From John Sirois <jsir...@apache.org>
Subject Review Request 42801: `TaskHistoryPruner` controls Lifecycle directly.
Date Tue, 26 Jan 2016 14:53:01 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/42801/
-----------------------------------------------------------

Review request for Aurora, Bill Farner and Zameer Manji.


Bugs: AURORA-1593
    https://issues.apache.org/jira/browse/AURORA-1593


Repository: aurora


Description
-------

This was the original idea in https://reviews.apache.org/r/42332.

Mixing the active scheduler `Service` lifecycle with the `EventBus`
lifecycle proves tricky - prune events are fired before scheduler active
services are started.  Instead of queueing up prune events to wait for
service start or re-engineering service / event bus interaction, returns
to the orignal behavior, manipulating the `Lifecycle` directly.

Also kill a confusing unused EventSink discovered during analyis of all
pub-sub event sourcing that might interact with the `TaskHistoryPruner`.

 src/main/java/org/apache/aurora/GuavaUtils.java                              | 18 ------------------
 src/main/java/org/apache/aurora/scheduler/SchedulerLifecycle.java            |  5 -----
 src/main/java/org/apache/aurora/scheduler/pruning/PruningModule.java         |  1 -
 src/main/java/org/apache/aurora/scheduler/pruning/TaskHistoryPruner.java     | 24 +++++++++++++-----------
 src/test/java/org/apache/aurora/LifecycleShutdownListenerTest.java           | 17 +++++++++++++++--
 src/test/java/org/apache/aurora/scheduler/SchedulerLifecycleTest.java        |  4 ----
 src/test/java/org/apache/aurora/scheduler/pruning/TaskHistoryPrunerTest.java | 24 ++++++++----------------
 7 files changed, 36 insertions(+), 57 deletions(-)


Diffs
-----

  src/main/java/org/apache/aurora/GuavaUtils.java 7d569e000c358e4f27397168843afdc0e9471ad6

  src/main/java/org/apache/aurora/scheduler/SchedulerLifecycle.java 5ba5e73464336014d5d49885a080dc98bebfeefb

  src/main/java/org/apache/aurora/scheduler/pruning/PruningModule.java efdfbdaa4ea00469faf2c33a8bc12051685f87ca

  src/main/java/org/apache/aurora/scheduler/pruning/TaskHistoryPruner.java 5441630522b3855a3b2036b5ff66fe980728ec68

  src/test/java/org/apache/aurora/LifecycleShutdownListenerTest.java d91dc277efebf658947382614f77aff15b3a251b

  src/test/java/org/apache/aurora/scheduler/SchedulerLifecycleTest.java bab45670d66dfed23dd6e0339687166c9be44ba4

  src/test/java/org/apache/aurora/scheduler/pruning/TaskHistoryPrunerTest.java ffeee1bebdb403be9179a8160e4d9a01aaf1f56b


Diff: https://reviews.apache.org/r/42801/diff/


Testing
-------

Locally green:
```
./gradlew -Pq build
./src/test/sh/org/apache/aurora/e2e/test_end_to_end.sh
```
It's the latter - e2e (krb part) - that was the only automated testing
revealing the problem previously.


Thanks,

John Sirois


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message