mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alexander Rukletsov <ruklet...@gmail.com>
Subject Review Request 63662: Dropped scheduler call if a connection has been interrupted.
Date Wed, 08 Nov 2017 11:04:17 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/63662/
-----------------------------------------------------------

Review request for mesos, Greg Mann and Till Toenshoff.


Bugs: MESOS-8179
    https://issues.apache.org/jira/browse/MESOS-8179


Repository: mesos


Description
-------

Prior to this patch, if a connection to the master is interrupted
between `send()` and `_send()`, executor library will crash on
`CHECK`. Since the connection can turn into pumpkin at any time, we
should not fail in this case.


Diffs
-----

  src/scheduler/scheduler.cpp fe374541621015d04a594f68d009ce50ec751d30 


Diff: https://reviews.apache.org/r/63662/diff/1/


Testing
-------

make check on Mac OS 10.11.6

Additionally run
```
GLOG_v=1 GTEST_FILTER="*SchedulerTest.MasterFailover*" ./bin/mesos-tests.sh --verbose --gtest_repeat=100
--gtest_break_on_failure
```
and grepped for "Connection to master" to ensure that this path is taken. `SchedulerTest.MasterFailover`
is modified locally to increase chances by adding multiple `detector->appoint(master.get()->pid);`
statements.


Thanks,

Alexander Rukletsov


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