Commit Graph

1735 Commits

Author SHA1 Message Date
Chris Douglas 2f8825588e YARN-5164. Use plan RLE to improve CapacityOverTimePolicy efficiency
(cherry picked from commit d383bfdcd4)
2017-09-08 14:44:34 -07:00
Arun Suresh 2b09cec218 YARN-6849. NMContainerStatus should have the Container ExecutionType. (Kartheek Muthyala via asuresh)
(cherry picked from commit 1f53ae7972)
2017-09-08 09:26:50 -07:00
Sunil G a39829c456 YARN-6600. Introduce default and max lifetime of application at LeafQueue level. Contributed by Rohith Sharma K S. 2017-09-08 19:20:52 +05:30
Sunil G ef342455cb YARN-6248. user is not removed from UsersManager’s when app is killed with pending container requests. Contributed by Eric Payne.
(cherry picked from commit c1386a90d7)
2017-09-07 14:47:07 -05:00
Subru Krishnan 5ccdd83923 YARN-5328. Plan/ResourceAllocation data structure enhancements required to support recurring reservations in ReservationSystem.
(cherry picked from commit b6e7d13696)
2017-09-06 16:46:49 -07:00
Wei-Chiu Chuang 92d9ad7761 Revert "Plan/ResourceAllocation data structure enhancements required to support recurring reservations in ReservationSystem."
This reverts commit 80516b3de7.
2017-09-05 14:52:43 -07:00
Subru Krishnan 80516b3de7 Plan/ResourceAllocation data structure enhancements required to support recurring reservations in ReservationSystem.
(cherry picked from commit 7996eca7dc)
2017-09-01 15:17:27 -07:00
Jason Lowe 57b5b186ce YARN-7120. CapacitySchedulerPage NPE in "Aggregate scheduler counts" section. Contributed by Eric Payne
(cherry picked from commit 99a7f5d451)
2017-09-01 11:10:40 -05:00
Daniel Templeton 004231dc08 YARN-6780. ResourceWeights.toString() cleanup (Contributed by weiyuan via Daniel Templeton)
(cherry picked from commit d4417dae4f)
2017-08-31 09:38:31 -07:00
Sunil G e8b4427259 YARN-7116. CapacityScheduler Web UI: Queue's AM usage is always show on per-user's AM usage. Contributed by Wangda Tan.
(cherry picked from commit f9e0cc8cdc)
2017-08-31 18:43:17 +05:30
Rohith Sharma K S 877d96b83d YARN-6982. Potential issue on setting AMContainerSpec#tokenConf to null before app is completed. Contributed by Manikandan R.
(cherry picked from commit 4cae120c61)
2017-08-30 10:50:57 +05:30
Jason Lowe c930aa3ff9 YARN-5816. TestDelegationTokenRenewer#testCancelWithMultipleAppSubmissions is still flakey. Contributed by Robert Kanter 2017-08-29 10:02:44 -05:00
Junping Du f252e1fe28 YARN-7076. yarn application -list -appTypes is not working. Contributed by Jian He.
(cherry picked from commit 312b1fd9da)
2017-08-28 18:27:38 -07:00
Sunil G 80570737a8 YARN-7051. Avoid concurrent modification exception in FifoIntraQueuePreemptionPlugin. Contributed by Eric Payne.
(cherry picked from commit 02599bda04)
2017-08-28 12:27:29 +05:30
Jason Lowe ea5bb5898f YARN-7052. RM SchedulingMonitor gives no indication why the spawned thread crashed. Contributed by Eric Payne
(cherry picked from commit 39a9dc8e4a)
2017-08-25 15:32:29 -05:00
Jason Lowe 82957ce0d0 YARN-6640. AM heartbeat stuck when responseId overflows MAX_INT. Contributed by Botong Huang
(cherry picked from commit 3a4e861169)
2017-08-25 09:20:50 -05:00
Yufei Gu 389b659c94 YARN-7049. FSAppAttempt preemption related fields have confusing names. (Contributed by Karthik Kambatla via Yufei Gu)
(cherry picked from commit 9e2699ac2c)
2017-08-24 17:54:00 -07:00
Junping Du c98917f798 YARN-7090. testRMRestartAfterNodeLabelDisabled get failed when CapacityScheduler is configured. Contributed by Wangda Tan.
(cherry picked from commit 652dd434d9)
2017-08-23 18:07:07 -07:00
Wangda Tan 245b49332d YARN-6251. Do async container release to prevent deadlock during container updates. (Arun Suresh via wangda)
Change-Id: I6c67d20c5dd4d22752830ebf0ed2340824976ecb
(cherry picked from commit f49843a988)
2017-08-23 10:35:37 -07:00
Subru Krishnan a45ffdcddc YARN-7053. Move curator transaction support to ZKCuratorManager. (Jonathan Hung via Subru).
(cherry picked from commit 4249172e14)
2017-08-22 19:21:35 -07:00
Jason Lowe a7893033fe YARN-7048. Fix tests faking kerberos to explicitly set ugi auth type. Contributed by Daryn Sharp
(cherry picked from commit 657dd59cc8)
2017-08-22 13:17:31 -05:00
Jason Lowe fc7df25a11 YARN-2416. InvalidStateTransitonException in ResourceManager if AMLauncher does not receive response for startContainers() call in time. Contributed by Jonathan Eagles
(cherry picked from commit 3efcd51c3b)
2017-08-22 12:59:47 -05:00
Arun Suresh f8f193c8e8 YARN-6979. Add flag to notify all types of container updates to NM via NodeHeartbeatResponse. (Kartheek Muthyala via asuresh)
(cherry picked from commit 8410d862d3)
2017-08-20 08:31:17 -07:00
Yufei Gu 4bd417a056 YARN-6969. Clean up unused code in class FairSchedulerQueueInfo. (Larry Lo via Yufei Gu)
(cherry picked from commit 8991f0baec)
2017-08-18 14:45:13 -07:00
Daniel Templeton 78f9dca2cb YARN-6952. Enable scheduling monitor in FS. (Yufei Gu via Daniel Templeton) 2017-08-18 14:40:30 -07:00
bibinchundatt ecf1a029bb YARN-7007. NPE in RM while using YarnClient.getApplications(). Contributed by Lingfeng Su.
(cherry picked from commit e05fa3451d)
2017-08-18 20:31:26 +05:30
Naganarasimha fa3094e436 YARN-6327. Removing queues from CapacitySchedulerQueueManager and ParentQueue should be done with iterator. Contributed by Jonathan Hung.
(cherry picked from commit 0a3aa40fe7)
2017-08-17 10:42:00 +05:30
bibinchundatt 67f9968407 YARN-6741. Deleting all children of a Parent Queue on refresh throws exception. Contributed by Naganarasimha G R. 2017-08-16 14:52:09 +05:30
Akira Ajisaka a28a3dc8f2
YARN-6965. Duplicate instantiation in FairSchedulerQueueInfo. Contributed by Masahiro Tanaka.
(cherry picked from commit 588c190afd)
2017-08-16 14:07:08 +09:00
Subru Krishnan ad46a90872 HADOOP-14773. Extend ZKCuratorManager API for more reusability. (Íñigo Goiri via Subru).
(cherry picked from commit 75dd866bfb)
2017-08-15 16:54:48 -07:00
Arun Suresh 7b22df3da6 YARN-5978. ContainerScheduler and ContainerManager changes to support ExecType update. (Kartheek Muthyala via asuresh)
(cherry picked from commit 4d7be1d857)
2017-08-14 19:57:34 -07:00
Jason Lowe 56d0e757c7 YARN-6917. Queue path is recomputed from scratch on every allocation. Contributed by Eric Payne
(cherry picked from commit 5558792894)
2017-08-14 15:34:50 -05:00
Jason Lowe e5ba40cc98 YARN-6987. Log app attempt during InvalidStateTransition. Contributed by Jonathan Eagles
(cherry picked from commit 3325ef653d)
2017-08-14 14:41:38 -05:00
Subru Krishnan a70efb6138 HADOOP-14741. Refactor curator based ZooKeeper communication into common library. (Íñigo Goiri via Subru). 2017-08-14 11:03:50 -07:00
Carlo Curino 394573780b YARN-6687. Validate that the duration of the periodic reservation is less than the periodicity. (subru via curino)
(cherry picked from commit 28d97b79b6)
2017-08-14 11:01:56 -07:00
Jian He 630ada1b4e YARN-6959. RM may allocate wrong AM Container for new attempt. Contributed by Yuqi Wang 2017-08-14 10:56:39 -07:00
Daniel Templeton 7d3cea232d Revert "YARN-6882. AllocationFileLoaderService.reloadAllocations() should use the diamond operator"
This reverts commit ce0cdc50d9.
2017-08-11 20:52:07 -07:00
Daniel Templeton ce0cdc50d9 YARN-6882. AllocationFileLoaderService.reloadAllocations() should use the diamond operator
(Contributed by Larry Lo via Daniel Templeton)

(cherry picked from commit 0996acde6c)
2017-08-11 14:51:42 -07:00
Daniel Templeton 11e3673748 YARN-6967. Limit application attempt's diagnostic message size thoroughly
(Contributed by Chengbing Liu via Daniel Templeton)

(cherry picked from commit 65364defb4)
2017-08-11 14:31:11 -07:00
Daniel Templeton 2566e5fcf7 YARN-6884. AllocationFileLoaderService.loadQueue() has an if without braces
(Contributed by weiyuan via Daniel Templeton)

(cherry picked from commit c7680d4cc4)
2017-08-11 14:23:17 -07:00
Subru Krishnan e9ec1e8f87 YARN-6897. Refactoring RMWebServices by moving some util methods to RMWebAppUtil. (Giovanni Matteo Fumarola via Subru). 2017-08-08 14:01:22 -07:00
Daniel Templeton c5b3d80a4b YARN-6879. TestLeafQueue.testDRFUserLimits() has commented out code
(Contributed by Angela Wang via Daniel Templeton)

(cherry picked from commit e0c24145d2)
2017-08-08 13:36:49 -07:00
Wangda Tan e18415a350 YARN-4161. Capacity Scheduler : Assign single or multiple containers per heart beat driven by configuration. (Wei Yan via wangda)
Change-Id: Ic441ae4e0bf72e7232411eb54243ec143d5fd0d3
(cherry picked from commit adb84f34db)
2017-08-07 11:34:45 -07:00
Yufei Gu b9549e108e YARN-6802. Add max AM resource and AM resource usage to leaf queue view in FairScheduler WebUI. (YunFan Zhou via Yufei Gu) 2017-08-04 17:23:44 -07:00
Carlo Curino 7654b52167 YARN-6634. [API] Refactor ResourceManager WebServices to make API explicit (Contributed by Giovanni Matteo Fumarola via curino).
(Edited cherry-pick of 547cb53e8a).
2017-08-04 17:04:20 -07:00
Arun Suresh 4cfd248a72 YARN-6777. Support for ApplicationMasterService processing chain of interceptors. (asuresh)
(cherry picked from commit 077fcf6a96)
2017-08-04 16:50:21 -07:00
Arun Suresh 043b7d133e YARN-6776. Refactor ApplicaitonMasterService to move actual processing logic to a separate class. (asuresh)
(cherry picked from commit 5496a34c0c)
2017-08-04 16:43:35 -07:00
Arun Suresh 2d62af6545 YARN-5966. AMRMClient changes to support ExecutionType update. (asuresh)
(cherry picked from commit aaf106fde3)
2017-08-04 16:19:44 -07:00
Arun Suresh 88d6adf419 YARN-5977. ContainerManagementProtocol changes to support change of container ExecutionType. (Kartheek Muthyala via asuresh)
(cherry picked from commit 35dc782923)
2017-08-03 21:18:43 -07:00
Yufei Gu b6729a7be5 YARN-6895. [FairScheduler] Preemption reservation may cause regular reservation leaks. (Miklos Szegedi via Yufei Gu) 2017-08-03 16:56:15 -07:00
Yufei Gu ce609d06c0 YARN-6832. Tests use assertTrue(....equals(...)) instead of assertEquals()). (Daniel Templeton via Yufei Gu) 2017-08-03 13:52:58 -07:00
Sunil G 82ec700241 YARN-6678. Handle IllegalStateException in Async Scheduling mode of CapacityScheduler. Contributed by Tao Yang.
(cherry picked from commit f64cfeaf61)
2017-08-03 19:28:16 +05:30
Eric Payne 223229b99c YARN-5349. TestWorkPreservingRMRestart#testUAMRecoveryOnRMWorkPreservingRestart fail intermittently. Contributed by Jason Lowe.
(cherry picked from commit 8ce8672b6b)
2017-08-02 12:58:07 -05:00
Jian He 6ee0fe70c7 YARN-6872. [Addendum patch] Ensure apps could run given NodeLabels are disabled post RM switchover/restart. Contributed by Sunil G 2017-08-02 00:03:59 -07:00
Jian He 4006cc266b YARN-6872. Ensure apps could run given NodeLabels are disabled post RM switchover/restart. Contributed by Sunil G 2017-08-01 10:00:04 -07:00
Yufei Gu e29d1c75ea YARN-6864. FSPreemptionThread cleanup for readability. (Daniel Templeton via Yufei Gu)
(cherry picked from commit 9902be72cb)
2017-07-27 23:21:30 -07:00
Varun Saxena 05bbbcbdbe Addendum for YARN-5548. Use MockRMMemoryStateStore to reduce test failures (Bibin A Chundatt via Varun Saxena)
(cherry picked from commit 11ece0bda1)
2017-07-28 00:33:41 +05:30
Varun Saxena 0b342a644a YARN-5548. Use MockRMMemoryStateStore to reduce test failures (Bibin A Chundatt via Varun Saxena)
(cherry picked from commit f66fd11e51)

Conflicts:
	hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/MockRM.java
	hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/TestRMRestart.java
2017-07-28 00:33:37 +05:30
Sunil G 08d5f3acf6 Addendum patch for YARN-6102. 2017-07-26 18:40:52 +05:30
Daniel Templeton 00f4a80e8f YARN-6307. Refactor FairShareComparator#compare (Contributed by Yufei Gu via Daniel Templeton)
(cherry picked from commit f81a4efb8c)
2017-07-25 13:02:00 -07:00
Akira Ajisaka 22f237c5d9
HADOOP-14681. Remove MockitoMaker class. Contributed by Andras Bokor.
(cherry picked from commit cca51e916b)
2017-07-25 15:25:36 +09:00
Daniel Templeton 9b6e787230 YARN-6845. Variable scheduler of FSLeafQueue duplicates the one of its parent FSQueue.
(Contributed by Yufei Gu via Daniel Templeton)

(cherry picked from commit 10583625c1)
2017-07-24 13:45:09 -07:00
Daniel Templeton eb1751670a YARN-6779. DominantResourceFairnessPolicy.DominantResourceFairnessComparator.calculateShares() should be @VisibleForTesting
(Contributed by Yeliang Cang via Daniel Templeton)

(cherry picked from commit bb30bd3771)
2017-07-24 12:15:35 -07:00
Sunil G a657472b42 YARN-6102. RMActiveService context to be updated with new RMContext on failover. Contributed by Rohith Sharma K S. 2017-07-24 20:57:56 +05:30
Wangda Tan 5070c9ba39 YARN-6775. CapacityScheduler: Improvements to assignContainers, avoid unnecessary canAssignToUser/Queue calls. (Nathan Roberts via wangda)
Change-Id: Iaf9bb7e5ed3aa1300abdccf87ae6fcbddbd25e3e
2017-07-19 15:25:21 -07:00
Rohith Sharma K S f806e8627b YARN-6819. Application report fails if app rejected due to nodesize. Contributed by Bibin A Chundatt.
(cherry picked from commit 845c4e52bd)
2017-07-19 11:16:17 +05:30
Yufei Gu d04bb465d2 YARN-6778. In ResourceWeights, weights and setWeights() should be final. (Daniel Templeton via Yufei Gu)
(cherry picked from commit daaf530fce)
2017-07-18 16:40:13 -07:00
Sunil G 487b0a3d7f Addendum patch for YARN-5731
(cherry picked from commit 0b7afc060c)
2017-07-18 13:46:43 +05:30
Sunil G 9b1abb448d YARN-6714. IllegalStateException while handling APP_ATTEMPT_REMOVED event when async-scheduling enabled in CapacityScheduler. Contributed by Tao Yang. 2017-07-17 13:22:04 +05:30
Sunil G d462e48336 YARN-6280. Introduce deselect query param to skip ResourceRequest from getApp/getApps REST API. Contributed by Lantao Jin. 2017-07-17 12:25:41 +05:30
Jason Lowe 9eb93c6f3f YARN-3260. AM attempt fail to register before RM processes launch event. Contributed by Bibin A Chundatt
(cherry picked from commit a5ae5ac50e)
2017-07-14 15:03:48 -05:00
bibinchundatt af80d10e54 YARN-6759. Fix TestRMRestart.testRMRestartWaitForPreviousAMToFinish failure. Contributed by Naganarasimha G R
(cherry picked from commit 75c0220b44)
2017-07-14 14:05:38 +05:30
Yufei Gu b2a34fe8d6 YARN-6769. Make schedulables without demand less needy in FairSharePolicy#compare. (Yunfan Zhou via Yufei Gu)
(cherry picked from commit 4a574e9a84)
2017-07-13 23:11:27 -07:00
Sunil G fd6935709f YARN-5731. Preemption calculation is not accurate when reserved containers are present in queue. Contributed by Wangda Tan. 2017-07-14 08:45:17 +05:30
Sunil G f5cdee6bfd YARN-6792. Incorrect XML convertion in NodeIDsInfo and LabelsToNodesInfo. Contributed by Giovanni Matteo Fumarola.
(cherry picked from commit 228ddaa31d)
2017-07-14 08:08:32 +05:30
Sunil G d83e8712f6 YARN-5892. Support user-specific minimum user limit percentage in Capacity Scheduler. Contributed by Eric Payne.
(cherry picked from commit ca13b224b2)
2017-07-13 14:50:28 -05:00
Wangda Tan cbb5f601f1 YARN-2113. Add cross-user preemption within CapacityScheduler's leaf-queue. (Sunil G via wangda)
Change-Id: I9b19f69788068be05b3295247cdd7b972f8a573c
(cherry picked from commit c583ab02c7)
2017-07-13 10:44:07 -05:00
Xuan edc0aa95f1 YARN-6689. PlacementRule should be configurable. (Jonathan Hung via xgong) 2017-07-12 15:45:51 -07:00
Wangda Tan f2d440b3b3 YARN-5889. Improve and refactor user-limit calculation in Capacity Scheduler. (Sunil G via wangda)
(cherry picked from commit 5fb723bb77)
2017-07-12 10:06:02 -05:00
Eric Payne e6cdf770ca Revert "YARN-2113. Add cross-user preemption within CapacityScheduler's leaf-queue. (Contributed by Sunil G)"
This reverts commit eda4ac07c1.
Commit eda4ac07c1 was a separate patch from trunk rather than a cherry-pick. I will cherryp-ick dependencies and then cherry-pick the trunk commit for YARN-2113.
2017-07-12 08:56:40 -05:00
Yufei Gu 20a2770d72 YARN-6764. Simplify the logic in FairScheduler#attemptScheduling. Contributed by Yufei Gu.
(cherry picked from commit 3de47ab5ea)
2017-07-09 16:11:26 -07:00
Naganarasimha b08ef44d9e YARN-6428. Queue AM limit is not honored in CS always. Contributed by Bibin A Chundatt.
(cherry picked from commit 626d730bfc)
2017-07-09 17:27:50 +05:30
Daniel Templeton fcef28e53b YARN-6410. FSContext.scheduler should be final (Contributed by Yeliang Cang via Daniel Templeton)
(cherry picked from commit 9d27880442)
2017-07-09 19:38:37 +09:00
Daniel Templeton a6d194d720 YARN-6746. SchedulerUtils.checkResourceRequestMatchingNodePartition() is dead code
(Contributed by Deepti Sawhney via Daniel Templeton)

(cherry picked from commit c5a0c38d8e)
2017-07-09 19:31:20 +09:00
Daniel Templeton edcfae8258 YARN-6752. Display reserved resources in web UI per application
(Contributed by Abdullah Yousufi via Daniel Templeton)

(cherry picked from commit 0615985886)
2017-07-09 19:19:07 +09:00
Eric Payne eda4ac07c1 YARN-2113. Add cross-user preemption within CapacityScheduler's leaf-queue. (Contributed by Sunil G) 2017-07-06 10:43:00 -05:00
Naganarasimha 87c81cfad3 YARN-6681. Eliminate double-copy of child queues in canAssignToThisQueue. Contributed by Daryn Sharp.
(cherry picked from commit fa1aaee87b)
2017-07-01 12:32:01 +05:30
Daniel Templeton 93c5421a43 YARN-6751. Display reserved resources in web UI per queue
(Contributed by Abdullah Yousufi via Daniel Templeton)

(cherry picked from commit ec97519779)
2017-06-29 16:57:23 -07:00
Naganarasimha a5ae7c0cae YARN-6467. CSQueueMetrics needs to update the current metrics for default partition only. Contributed by Manikandan R. 2017-06-29 00:09:30 +05:30
Naganarasimha 92e266c2b6 YARN-5006. ResourceManager quit due to ApplicationStateData exceed the limit size of znode in zk. Contributed by Bibin A Chundatt. 2017-06-24 23:14:51 +05:30
Yufei Gu 4c59b446e2 YARN-5876. TestResourceTrackerService#testGracefulDecommissionWithApp fails intermittently on trunk. (Robert Kanter via Yufei Gu)
(cherry picked from commit 0b77262890)
2017-06-23 13:30:20 -07:00
Akira Ajisaka 35f6c7c9c7
YARN-6713. Fix dead link in the Javadoc of FairSchedulerEventLog.java. Contributed by Weiwei Yang.
(cherry picked from commit 099dfe92a9)
2017-06-20 17:23:06 +09:00
Naganarasimha 578318473c YARN-6680. Avoid locking overhead for NO_LABEL lookups. Contributed by Daryn Sharp
(cherry picked from commit ee89ac84e6)
2017-06-19 22:45:21 +05:30
bibinchundatt 3f9acf65fb YARN-6703. RM startup failure with old state store due to version mismatch. Contributed by Varun Saxena
(cherry picked from commit d64c842743)
2017-06-12 18:32:23 +05:30
Eric Payne 33605e0896 YARN-6585. RM fails to start when upgrading from 2.7 for clusters with node labels. Contributed by Sunil G.
(cherry picked from commit 5578af8603)
2017-06-09 16:18:16 -05:00
Naganarasimha 85d561a5b3 YARN-6681. Improve performance of AssignmentInformation datastructures. Contributed by Daryn sharp. 2017-06-09 11:46:51 +05:30
Daniel Templeton 10b7a5fa98 YARN-6246. Identifying starved apps does not need the scheduler writelock
(Contributed by Karthik Kambatla via Daniel Templeton)

(cherry picked from commit d5b71e4175)
2017-05-31 15:50:11 -07:00
Jonathan Eagles f5f12b576e YARN-6497. Method length of ResourceManager#serviceInit() is too long (Gergely Novák via jeagles)
(cherry picked from commit cbfed0e82f)
2017-05-31 10:19:51 -05:00
Yufei Gu fc6cb4b2dd YARN-6582. FSAppAttempt demand can be updated atomically in updateDemand(). (Karthik Kambatla via Yufei Gu)
(cherry picked from commit 87590090c8)
2017-05-25 14:25:49 -07:00
Jason Lowe 4c44ff69df YARN-6643. TestRMFailover fails rarely due to port conflict. Contributed by Robert Kanter
(cherry picked from commit 3fd6a2da4e)
2017-05-25 16:09:08 -05:00