-
Notifications
You must be signed in to change notification settings - Fork 28.7k
[SPARK-5338][MESOS] Add cluster mode support for Mesos #5144
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
@pwendell @andrewor14 Let me know what you think! |
Test build #29030 has started for PR 5144 at commit
|
Test build #29030 has finished for PR 5144 at commit
|
Test FAILed. |
Test build #29038 has started for PR 5144 at commit
|
Test build #29038 has finished for PR 5144 at commit
|
Test FAILed. |
Test build #29040 has started for PR 5144 at commit
|
Test build #29040 has finished for PR 5144 at commit
|
Test FAILed. |
Test build #29041 has started for PR 5144 at commit
|
Test build #29041 has finished for PR 5144 at commit
|
Test FAILed. |
Test build #29123 has started for PR 5144 at commit
|
Test build #29123 has finished for PR 5144 at commit
|
Test FAILed. |
2f66466
to
9ab138f
Compare
Test build #29155 has started for PR 5144 at commit
|
Test build #29155 has finished for PR 5144 at commit
|
Test PASSed. |
printErrorAndExit("Cluster deploy mode is currently not supported for Mesos clusters.") | ||
case (MESOS, CLUSTER) if args.isPython => | ||
printErrorAndExit("Cluster deploy mode is currently not supported for python " + | ||
"on Mesos clusters.") |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
for python applications on Mesos clusters
(to be consistent with the warning message for standalone cluster python)
38e6c4b
to
069e946
Compare
Test build #30718 has started for PR 5144 at commit |
Test build #30718 has finished for PR 5144 at commit
|
Test FAILed. |
Another unrelated test failure |
Test build #31043 has started for PR 5144 at commit |
retest this please |
Merged build triggered. |
Merged build started. |
Test build #31163 has started for PR 5144 at commit |
new MesosStatusRequestServlet(scheduler, masterConf) | ||
} | ||
|
||
private[deploy] class MesosSubmitRequestServlet( |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
why are these private[deploy]
? Can this be tighter (e.g. private)?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
never mind, otherwise we they would escape the scope
Hey @tnachen just wanted to let you know I just merged this into master after fixing the minor typos myself. Congratulations! Thanks for patiently re-iterating so many times on the code reviews. |
Test build #31163 has finished for PR 5144 at commit
|
Merged build finished. Test FAILed. |
Test FAILed. |
…r mode when HA is enabled **3/26 update:** * Akka-based: Use an array of `ActorSelection` to represent multiple master. Add an `activeMasterActor` for query status of driver. And will add lost masters( including the standby one) to `lostMasters`. When size of `lostMasters` equals or greater than # of all masters, we should give an error that all masters are not avalible. * Rest-based: When all masters are not available(throw an exception), we use akka gateway to submit apps. I have tested simply on standalone HA cluster(with two masters alive and one alive/one dead), it worked. There might remains some issues on style or message print, but we can check the solution then fix them together. /cc srowen andrewor14 Author: WangTaoTheTonic <[email protected]> Closes #5116 from WangTaoTheTonic/SPARK-6443 and squashes the following commits: 2a28aab [WangTaoTheTonic] based the newest change #5144 76fd411 [WangTaoTheTonic] rebase f4f972b [WangTaoTheTonic] rebase...again a41de0b [WangTaoTheTonic] rebase 220cb3c [WangTaoTheTonic] move connect exception inside 35119a0 [WangTaoTheTonic] style and compile issues 9d636be [WangTaoTheTonic] per Andrew's comments 979760c [WangTaoTheTonic] rebase e4f4ece [WangTaoTheTonic] fix failed test 5d23958 [WangTaoTheTonic] refact some duplicated code, style and comments 7a881b3 [WangTaoTheTonic] when one of masters is gone, we still can submit 2b011c9 [WangTaoTheTonic] fix broken tests 60d97a4 [WangTaoTheTonic] rebase fa1fa80 [WangTaoTheTonic] submit app to HA cluster in standalone cluster mode
This patch adds the support for cluster mode to run on Mesos. It introduces a new Mesos framework dedicated to launch new apps/drivers, and can be called with the spark-submit script and specifying --master flag to the cluster mode REST interface instead of Mesos master. Example: ./bin/spark-submit --deploy-mode cluster --class org.apache.spark.examples.SparkPi --master mesos://10.0.0.206:8077 --executor-memory 1G --total-executor-cores 100 examples/target/spark-examples_2.10-1.3.0-SNAPSHOT.jar 30 Part of this patch is also to abstract the StandaloneRestServer so it can have different implementations of the REST endpoints. Features of the cluster mode in this PR: - Supports supervise mode where scheduler will keep trying to reschedule exited job. - Adds a new UI for the cluster mode scheduler to see all the running jobs, finished jobs, and supervise jobs waiting to be retried - Supports state persistence to ZK, so when the cluster scheduler fails over it can pick up all the queued and running jobs Author: Timothy Chen <[email protected]> Author: Luc Bourlier <[email protected]> Closes apache#5144 from tnachen/mesos_cluster_mode and squashes the following commits: 069e946 [Timothy Chen] Fix rebase. e24b512 [Timothy Chen] Persist submitted driver. 390c491 [Timothy Chen] Fix zk conf key for mesos zk engine. e324ac1 [Timothy Chen] Fix merge. fd5259d [Timothy Chen] Address review comments. 1553230 [Timothy Chen] Address review comments. c6c6b73 [Timothy Chen] Pass spark properties to mesos cluster tasks. f7d8046 [Timothy Chen] Change app name to spark cluster. 17f93a2 [Timothy Chen] Fix head of line blocking in scheduling drivers. 6ff8e5c [Timothy Chen] Address comments and add logging. df355cd [Timothy Chen] Add metrics to mesos cluster scheduler. 20f7284 [Timothy Chen] Address review comments 7252612 [Timothy Chen] Fix tests. a46ad66 [Timothy Chen] Allow zk cli param override. 920fc4b [Timothy Chen] Fix scala style issues. 862b5b5 [Timothy Chen] Support asking driver status when it's retrying. 7f214c2 [Timothy Chen] Fix RetryState visibility e0f33f7 [Timothy Chen] Add supervise support and persist retries. 371ce65 [Timothy Chen] Handle cluster mode recovery and state persistence. 3d4dfa1 [Luc Bourlier] Adds support to kill submissions febfaba [Timothy Chen] Bound the finished drivers in memory 543a98d [Timothy Chen] Schedule multiple jobs 6887e5e [Timothy Chen] Support looking at SPARK_EXECUTOR_URI env variable in schedulers 8ec76bc [Timothy Chen] Fix Mesos dispatcher UI. d57d77d [Timothy Chen] Add documentation 825afa0 [Luc Bourlier] Supports more spark-submit parameters b8e7181 [Luc Bourlier] Adds a shutdown latch to keep the deamon running 0fa7780 [Luc Bourlier] Launch task through the mesos scheduler 5b7a12b [Timothy Chen] WIP: Making a cluster mode a mesos framework. 4b2f5ef [Timothy Chen] Specify user jar in command to be replaced with local. e775001 [Timothy Chen] Support fetching remote uris in driver runner. 7179495 [Timothy Chen] Change Driver page output and add logging 880bc27 [Timothy Chen] Add Mesos Cluster UI to display driver results 9986731 [Timothy Chen] Kill drivers when shutdown 67cbc18 [Timothy Chen] Rename StandaloneRestClient to RestClient and add sbin scripts e3facdd [Timothy Chen] Add Mesos Cluster dispatcher
…r mode when HA is enabled **3/26 update:** * Akka-based: Use an array of `ActorSelection` to represent multiple master. Add an `activeMasterActor` for query status of driver. And will add lost masters( including the standby one) to `lostMasters`. When size of `lostMasters` equals or greater than # of all masters, we should give an error that all masters are not avalible. * Rest-based: When all masters are not available(throw an exception), we use akka gateway to submit apps. I have tested simply on standalone HA cluster(with two masters alive and one alive/one dead), it worked. There might remains some issues on style or message print, but we can check the solution then fix them together. /cc srowen andrewor14 Author: WangTaoTheTonic <[email protected]> Closes apache#5116 from WangTaoTheTonic/SPARK-6443 and squashes the following commits: 2a28aab [WangTaoTheTonic] based the newest change apache#5144 76fd411 [WangTaoTheTonic] rebase f4f972b [WangTaoTheTonic] rebase...again a41de0b [WangTaoTheTonic] rebase 220cb3c [WangTaoTheTonic] move connect exception inside 35119a0 [WangTaoTheTonic] style and compile issues 9d636be [WangTaoTheTonic] per Andrew's comments 979760c [WangTaoTheTonic] rebase e4f4ece [WangTaoTheTonic] fix failed test 5d23958 [WangTaoTheTonic] refact some duplicated code, style and comments 7a881b3 [WangTaoTheTonic] when one of masters is gone, we still can submit 2b011c9 [WangTaoTheTonic] fix broken tests 60d97a4 [WangTaoTheTonic] rebase fa1fa80 [WangTaoTheTonic] submit app to HA cluster in standalone cluster mode
…r mode when HA is enabled **3/26 update:** * Akka-based: Use an array of `ActorSelection` to represent multiple master. Add an `activeMasterActor` for query status of driver. And will add lost masters( including the standby one) to `lostMasters`. When size of `lostMasters` equals or greater than # of all masters, we should give an error that all masters are not avalible. * Rest-based: When all masters are not available(throw an exception), we use akka gateway to submit apps. I have tested simply on standalone HA cluster(with two masters alive and one alive/one dead), it worked. There might remains some issues on style or message print, but we can check the solution then fix them together. /cc srowen andrewor14 Author: WangTaoTheTonic <[email protected]> Closes apache#5116 from WangTaoTheTonic/SPARK-6443 and squashes the following commits: 2a28aab [WangTaoTheTonic] based the newest change apache#5144 76fd411 [WangTaoTheTonic] rebase f4f972b [WangTaoTheTonic] rebase...again a41de0b [WangTaoTheTonic] rebase 220cb3c [WangTaoTheTonic] move connect exception inside 35119a0 [WangTaoTheTonic] style and compile issues 9d636be [WangTaoTheTonic] per Andrew's comments 979760c [WangTaoTheTonic] rebase e4f4ece [WangTaoTheTonic] fix failed test 5d23958 [WangTaoTheTonic] refact some duplicated code, style and comments 7a881b3 [WangTaoTheTonic] when one of masters is gone, we still can submit 2b011c9 [WangTaoTheTonic] fix broken tests 60d97a4 [WangTaoTheTonic] rebase fa1fa80 [WangTaoTheTonic] submit app to HA cluster in standalone cluster mode
This patch adds the support for cluster mode to run on Mesos. It introduces a new Mesos framework dedicated to launch new apps/drivers, and can be called with the spark-submit script and specifying --master flag to the cluster mode REST interface instead of Mesos master. Example: ./bin/spark-submit --deploy-mode cluster --class org.apache.spark.examples.SparkPi --master mesos://10.0.0.206:8077 --executor-memory 1G --total-executor-cores 100 examples/target/spark-examples_2.10-1.3.0-SNAPSHOT.jar 30 Part of this patch is also to abstract the StandaloneRestServer so it can have different implementations of the REST endpoints. Features of the cluster mode in this PR: - Supports supervise mode where scheduler will keep trying to reschedule exited job. - Adds a new UI for the cluster mode scheduler to see all the running jobs, finished jobs, and supervise jobs waiting to be retried - Supports state persistence to ZK, so when the cluster scheduler fails over it can pick up all the queued and running jobs Author: Timothy Chen <[email protected]> Author: Luc Bourlier <[email protected]> Closes apache#5144 from tnachen/mesos_cluster_mode and squashes the following commits: 069e946 [Timothy Chen] Fix rebase. e24b512 [Timothy Chen] Persist submitted driver. 390c491 [Timothy Chen] Fix zk conf key for mesos zk engine. e324ac1 [Timothy Chen] Fix merge. fd5259d [Timothy Chen] Address review comments. 1553230 [Timothy Chen] Address review comments. c6c6b73 [Timothy Chen] Pass spark properties to mesos cluster tasks. f7d8046 [Timothy Chen] Change app name to spark cluster. 17f93a2 [Timothy Chen] Fix head of line blocking in scheduling drivers. 6ff8e5c [Timothy Chen] Address comments and add logging. df355cd [Timothy Chen] Add metrics to mesos cluster scheduler. 20f7284 [Timothy Chen] Address review comments 7252612 [Timothy Chen] Fix tests. a46ad66 [Timothy Chen] Allow zk cli param override. 920fc4b [Timothy Chen] Fix scala style issues. 862b5b5 [Timothy Chen] Support asking driver status when it's retrying. 7f214c2 [Timothy Chen] Fix RetryState visibility e0f33f7 [Timothy Chen] Add supervise support and persist retries. 371ce65 [Timothy Chen] Handle cluster mode recovery and state persistence. 3d4dfa1 [Luc Bourlier] Adds support to kill submissions febfaba [Timothy Chen] Bound the finished drivers in memory 543a98d [Timothy Chen] Schedule multiple jobs 6887e5e [Timothy Chen] Support looking at SPARK_EXECUTOR_URI env variable in schedulers 8ec76bc [Timothy Chen] Fix Mesos dispatcher UI. d57d77d [Timothy Chen] Add documentation 825afa0 [Luc Bourlier] Supports more spark-submit parameters b8e7181 [Luc Bourlier] Adds a shutdown latch to keep the deamon running 0fa7780 [Luc Bourlier] Launch task through the mesos scheduler 5b7a12b [Timothy Chen] WIP: Making a cluster mode a mesos framework. 4b2f5ef [Timothy Chen] Specify user jar in command to be replaced with local. e775001 [Timothy Chen] Support fetching remote uris in driver runner. 7179495 [Timothy Chen] Change Driver page output and add logging 880bc27 [Timothy Chen] Add Mesos Cluster UI to display driver results 9986731 [Timothy Chen] Kill drivers when shutdown 67cbc18 [Timothy Chen] Rename StandaloneRestClient to RestClient and add sbin scripts e3facdd [Timothy Chen] Add Mesos Cluster dispatcher
…r mode when HA is enabled **3/26 update:** * Akka-based: Use an array of `ActorSelection` to represent multiple master. Add an `activeMasterActor` for query status of driver. And will add lost masters( including the standby one) to `lostMasters`. When size of `lostMasters` equals or greater than # of all masters, we should give an error that all masters are not avalible. * Rest-based: When all masters are not available(throw an exception), we use akka gateway to submit apps. I have tested simply on standalone HA cluster(with two masters alive and one alive/one dead), it worked. There might remains some issues on style or message print, but we can check the solution then fix them together. /cc srowen andrewor14 Author: WangTaoTheTonic <[email protected]> Closes apache#5116 from WangTaoTheTonic/SPARK-6443 and squashes the following commits: 2a28aab [WangTaoTheTonic] based the newest change apache#5144 76fd411 [WangTaoTheTonic] rebase f4f972b [WangTaoTheTonic] rebase...again a41de0b [WangTaoTheTonic] rebase 220cb3c [WangTaoTheTonic] move connect exception inside 35119a0 [WangTaoTheTonic] style and compile issues 9d636be [WangTaoTheTonic] per Andrew's comments 979760c [WangTaoTheTonic] rebase e4f4ece [WangTaoTheTonic] fix failed test 5d23958 [WangTaoTheTonic] refact some duplicated code, style and comments 7a881b3 [WangTaoTheTonic] when one of masters is gone, we still can submit 2b011c9 [WangTaoTheTonic] fix broken tests 60d97a4 [WangTaoTheTonic] rebase fa1fa80 [WangTaoTheTonic] submit app to HA cluster in standalone cluster mode
This patch adds the support for cluster mode to run on Mesos.
It introduces a new Mesos framework dedicated to launch new apps/drivers, and can be called with the spark-submit script and specifying --master flag to the cluster mode REST interface instead of Mesos master.
Example:
./bin/spark-submit --deploy-mode cluster --class org.apache.spark.examples.SparkPi --master mesos://10.0.0.206:8077 --executor-memory 1G --total-executor-cores 100 examples/target/spark-examples_2.10-1.3.0-SNAPSHOT.jar 30
Part of this patch is also to abstract the StandaloneRestServer so it can have different implementations of the REST endpoints.
Features of the cluster mode in this PR: