-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/7746/
-----------------------------------------------------------
(Updated Oct. 29, 2012, 7:34 p.m.)
Review request for mesos, Benjamin Hindman and Ben Mahler.
Changes
-------
BenM's comments.
Rebased off latest trunk
Description (updated)
-------
BenM's comments
Fix for master detector
This addresses bug MESOS-299.
https://issues.apache.org/jira/browse/MESOS-299
Diffs (updated)
-----
src/detector/detector.hpp d859b080b99e23d511458a27272db33c5486bb4b
src/detector/detector.cpp 62df8bdf539eb13b2a6dc00eb2f6a07381d59106
src/slave/slave.cpp 0321bc516166aacfd261c48f1f4293622d18ae0e
src/tests/zookeeper_test_server.hpp 06320439b993f9612ea01303f7446dadf97dc045
src/tests/zookeeper_tests.cpp 3f001affe0dd4b8002e99a658c47b8ea86ddb7d6
Diff: https://reviews.apache.org/r/7746/diff/
Testing
-------
Test output before the fix:
[vinod@VKone ~/workspace/apache/mesos/build (vinod/master_detector_fix)]$
GLOG_v=1 ./bin/mesos-tests.sh --verbose
--gtest_filter="*ZooKeeperTest.MasterDetectorExpire*"
....
....
I1026 00:05:39.087263 27279360 detector.cpp:286] Master detector connected to
ZooKeeper ...
I1026 00:05:39.087425 27279360 detector.cpp:303] Trying to create path '/mesos'
in ZooKeeper
I1026 00:05:39.087811 26742784 detector.cpp:286] Master detector connected to
ZooKeeper ...
I1026 00:05:39.087836 26742784 detector.cpp:303] Trying to create path '/mesos'
in ZooKeeper
I1026 00:05:39.102313 27279360 detector.cpp:467] Master detector found 0
registered masters
I1026 00:05:39.110910 26742784 detector.cpp:333] Created ephemeral/sequence
znode at '/mesos/0000000000'
I1026 00:05:39.111507 27815936 detector.cpp:467] Master detector found 1
registered masters
I1026 00:05:39.111590 26742784 detector.cpp:467] Master detector found 1
registered masters
I1026 00:05:39.114651 27815936 detector.cpp:502] Master detector got new master
pid: (1)@192.168.1.127:51375
I1026 00:05:39.114917 26742784 detector.cpp:502] Master detector got new master
pid: (1)@192.168.1.127:51375
2012-10-26
00:05:39,116:15851(0x10fa9f000):ZOO_ERROR@handle_socket_error_msg@1603: Socket
[127.0.0.1:51378] zk retcode=-4, errno=64(Host is down): failed while receiving
a server response
I1026 00:05:39.116739 27279360 detector.cpp:378] Master detector lost
connection to ZooKeeper, attempting to reconnect ...
2012-10-26
00:05:42,450:15851(0x10fa9f000):ZOO_ERROR@handle_socket_error_msg@1621: Socket
[127.0.0.1:51378] zk retcode=-112, errno=70(Stale NFS file handle):
sessionId=0x13a9be3f5b90000 has expired.
W1026 00:05:42.450742 27279360 detector.cpp:397] Master detector ZooKeeper
session expired!
I1026 00:05:42.454856 27279360 detector.cpp:286] Master detector connected to
ZooKeeper ...
I1026 00:05:42.454888 27279360 detector.cpp:303] Trying to create path '/mesos'
in ZooKeeper
I1026 00:05:42.501096 27279360 detector.cpp:467] Master detector found 1
registered masters
../../src/tests/zookeeper_tests.cpp:332: Failure
Failed
Waited too long for 'newMasterDetectedCall2'
../../src/tests/zookeeper_tests.cpp:324: Failure
Actual function call count doesn't match EXPECT_CALL(slave,
newMasterDetected(master.self()))...
Expected: to be called once
Actual: never called - unsatisfied and active
I1026 00:05:44.844130 1898458304 zookeeper_server.cpp:181] Shutdown
ZooKeeperServer on port 51378
[ FAILED ] ZooKeeperTest.MasterDetectorExpireZKSession (5929 ms)
[----------] 1 test from ZooKeeperTest (5929 ms total)
[----------] Global test environment tear-down
[==========] 1 test from 1 test case ran. (6147 ms total)
[ PASSED ] 0 tests.
[ FAILED ] 1 test, listed below:
[ FAILED ] ZooKeeperTest.MasterDetectorExpireZKSession
Test output after the fix:
[vinod@VKone ~/workspace/apache/mesos/build (vinod/master_detector_fix)]$
GLOG_v=1 ./bin/mesos-tests.sh --verbose
--gtest_filter="*ZooKeeperTest.MasterDetectorExpire*"
...
...
I1025 23:42:01.587967 26742784 detector.cpp:286] Master detector connected to
ZooKeeper ...
I1025 23:42:01.588099 26742784 detector.cpp:303] Trying to create path '/mesos'
in ZooKeeper
I1025 23:42:01.588544 27279360 detector.cpp:286] Master detector connected to
ZooKeeper ...
I1025 23:42:01.588577 27279360 detector.cpp:303] Trying to create path '/mesos'
in ZooKeeper
I1025 23:42:01.609194 26742784 detector.cpp:333] Created ephemeral/sequence
znode at '/mesos/0000000000'
I1025 23:42:01.610599 27279360 detector.cpp:467] Master detector found 1
registered masters
I1025 23:42:01.610780 26742784 detector.cpp:467] Master detector found 1
registered masters
I1025 23:42:01.613991 27279360 detector.cpp:502] Master detector got new master
pid: (1)@192.168.1.127:51025
I1025 23:42:01.614141 26742784 detector.cpp:502] Master detector got new master
pid: (1)@192.168.1.127:51025
2012-10-25
23:42:01,616:12480(0x10faa8000):ZOO_ERROR@handle_socket_error_msg@1603: Socket
[127.0.0.1:51028] zk retcode=-4, errno=64(Host is down): failed while receiving
a server response
I1025 23:42:01.616317 26742784 detector.cpp:378] Master detector lost
connection to ZooKeeper, attempting to reconnect ...
2012-10-25
23:42:04,950:12480(0x10faa8000):ZOO_ERROR@handle_socket_error_msg@1621: Socket
[127.0.0.1:51028] zk retcode=-112, errno=70(Stale NFS file handle):
sessionId=0x13a9bce54800001 has expired.
W1025 23:42:04.950316 26742784 detector.cpp:397] Master detector ZooKeeper
session expired!
I1025 23:42:04.954572 27815936 detector.cpp:286] Master detector connected to
ZooKeeper ...
I1025 23:42:04.954607 27815936 detector.cpp:303] Trying to create path '/mesos'
in ZooKeeper
I1025 23:42:05.008098 27815936 detector.cpp:467] Master detector found 1
registered masters
I1025 23:42:05.008566 27815936 detector.cpp:502] Master detector got new master
pid: (1)@192.168.1.127:51025
I1025 23:42:05.010418 1898458304 zookeeper_server.cpp:181] Shutdown
ZooKeeperServer on port 51028
[ OK ] ZooKeeperTest.MasterDetectorExpireZKSession (3633 ms)
[----------] 1 test from ZooKeeperTest (3634 ms total)
[----------] Global test environment tear-down
[==========] 1 test from 1 test case ran. (4068 ms total)
[ PASSED ] 1 test.
Thanks,
Vinod Kone
