watch should handle etcd old version exception #1075
Labels
Comments
|
cc @foxish |
|
We run into the same issue, any progress on this? |
|
@yujiantao For a simple fix, you can try comment out these lines https://github.com/fabric8io/kubernetes-client/blob/v4.0.5/kubernetes-client/src/main/java/io/fabric8/kubernetes/client/dsl/internal/WatchConnectionManager.java#L141-L143 |
|
We also running in the same issue with Spark, would be great to see the fix eventually |
|
This issue has been automatically marked as stale because it has not had any activity since 90 days. It will be closed if no further activity occurs within 7 days. Thank you for your contributions! |
|
It is still relevant for people using k8s-as-a-service on Azure. We applied the workaround mentioned by @chenchun and it works fine so far... |
|
This issue has been automatically marked as stale because it has not had any activity since 90 days. It will be closed if no further activity occurs within 7 days. Thank you for your contributions! |
|
This issue has been automatically marked as stale because it has not had any activity since 90 days. It will be closed if no further activity occurs within 7 days. Thank you for your contributions! |
@chenchun Is this something we maybe could put into the client? That for some watches you don't care about version problems. |
|
It is still relevant and requires a rebuild of Spark from the sources just for using hacked kubernetes-client 😞 |
|
@SergSlipushenko : Actually we added it in past but we had to remove it, see #1800 |
|
@stijndehaes I took a look at #1800, is it better to add a bool flag of whether or not do re-watching automatically when receive a version change? So that we won't break the contract of sending HTTP_GONE if resource version is old and also makes people easier when they don't care about the problem. |
|
@chenchun : Adding a boolean flag to reconnect sounds nice to me 👍 . This thing gets requested quite often. I think we need to tweak WatchConnectionManager |
rohanKanojia
added
enhancement
component/kubernetes-client
|
I also noticed there is a deprecated watch method method that allows you to set a resource version. Looking in the git history does not tell me much. But using that method allows you to set a null resource version, that way you don't get the HTTP GONE message I believe? |
|
We implemented SharedInformers (#1384) a while back to mimic client-go's behavior and provide an extra level of abstraction for Watch operations (Kubernetes client-go: watch.Interface vs. cache.NewInformer vs. cache.NewSharedIndexInformer? and Writing Controllers/SharedInformers) Our implementation of SharedInformers already takes care of HTTP_GONE scenario. If you are looking for this reconnect behavior, I would encourage using SharedInformers instead of Watch, or else use watch with your own reconnect implementation. I think providing this behavior for watch too would be duplicating a feature that's already available in Informers. @rohanKanojia maybe we can use this issue to provide some additional examples and documentation on different use-cases for SharedInformers. I think it's unclear that they should be the default approach to watch resources. |
|
@manusa one big difference is that with a watcher we can watch one single pod. This is watch spark-submit does when watching the driver, with sharedinformer I am watching all the pods. Unless there is way to watch a single pod? |
|
I'm really unsure how we implemented the SharedInformer and what are the current features, but there should be an option to filter by labels (or even fields>i.e. metadata.name). |
|
@manusa found it! You can do it like this I think: |
|
🎉 Good news @stijndehaes, thx for sharing! |
|
@stijndehaes : Hi, I tried documenting On debugging, I realized that we were losing [0] https://medium.com/@rohaan/introduction-to-fabric8-kubernetes-java-client-informer-api-b945082d69af |
|
With my PR, you should be able to get Upon debugging I checked that when we query a single resource response is in the form of a single resource not in the form of a list. Hence, Deserialization fails during list step resulting in resource's I checked client-go's implementation but I'm not sure if they support listing a specific resource. Maybe informers are not meant to list specific resources? Since they are implementing |
asfgit
pushed a commit
to apache/spark
that referenced
this issue
Jul 21, 2020
…ged from k8s ### What changes were proposed in this pull request? Restart the watcher when it failed with a HTTP_GONE code from the kubernetes api. Which means a resource version has changed. For more relevant information see here: fabric8io/kubernetes-client#1075 ### Why are the changes needed? ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? Running spark-submit to a k8s cluster. Not sure how to make an automated test for this. If someone can help me out that would be great. Closes #28423 from stijndehaes/bugfix/k8s-submit-resource-version-change. Authored-by: Stijn De Haes <[email protected]> Signed-off-by: Holden Karau <[email protected]>
|
This issue has been automatically marked as stale because it has not had any activity since 90 days. It will be closed if no further activity occurs within 7 days. Thank you for your contributions! |
jkleckner
pushed a commit
to jkleckner/spark
that referenced
this issue
Aug 30, 2020
…ged from k8s ### What changes were proposed in this pull request? Restart the watcher when it failed with a HTTP_GONE code from the kubernetes api. Which means a resource version has changed. For more relevant information see here: fabric8io/kubernetes-client#1075 ### Why are the changes needed? ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? Running spark-submit to a k8s cluster. Not sure how to make an automated test for this. If someone can help me out that would be great. Closes apache#28423 from stijndehaes/bugfix/k8s-submit-resource-version-change. Authored-by: Stijn De Haes <[email protected]> Signed-off-by: Holden Karau <[email protected]>
jkleckner
pushed a commit
to jkleckner/spark
that referenced
this issue
Sep 25, 2020
…ged from k8s Restart the watcher when it failed with a HTTP_GONE code from the kubernetes api. Which means a resource version has changed. For more relevant information see here: fabric8io/kubernetes-client#1075 No Running spark-submit to a k8s cluster. Not sure how to make an automated test for this. If someone can help me out that would be great. Closes apache#28423 from stijndehaes/bugfix/k8s-submit-resource-version-change. Address review comment to fully qualify import scala.util.control
jkleckner
pushed a commit
to jkleckner/spark
that referenced
this issue
Sep 25, 2020
…ged from k8s Restart the watcher when it failed with a HTTP_GONE code from the kubernetes api. Which means a resource version has changed. For more relevant information see here: fabric8io/kubernetes-client#1075 No Running spark-submit to a k8s cluster. Not sure how to make an automated test for this. If someone can help me out that would be great. Closes apache#28423 from stijndehaes/bugfix/k8s-submit-resource-version-change. Address review comment to fully qualify import scala.util.control
jkleckner
pushed a commit
to jkleckner/spark
that referenced
this issue
Sep 28, 2020
…ged from k8s Restart the watcher when it failed with a HTTP_GONE code from the kubernetes api. Which means a resource version has changed. For more relevant information see here: fabric8io/kubernetes-client#1075 No Running spark-submit to a k8s cluster. Not sure how to make an automated test for this. If someone can help me out that would be great. Closes apache#28423 from stijndehaes/bugfix/k8s-submit-resource-version-change. Address review comment to fully qualify import scala.util.control Rebase on branch-3.0 to fix SparkR integration test.
holdenk
added a commit
to holdenk/spark
that referenced
this issue
Oct 27, 2020
[SPARK-21040][CORE] Speculate tasks which are running on decommission executors This PR adds functionality to consider the running tasks on decommission executors based on some config. In spark-on-cloud , we sometimes already know that an executor won't be alive for more than fix amount of time. Ex- In AWS Spot nodes, once we get the notification, we know that a node will be gone in 120 seconds. So if the running tasks on the decommissioning executors may run beyond currentTime+120 seconds, then they are candidate for speculation. Currently when an executor is decommission, we stop scheduling new tasks on those executors but the already running tasks keeps on running on them. Based on the cloud, we might know beforehand that an executor won't be alive for more than a preconfigured time. Different cloud providers gives different timeouts before they take away the nodes. For Ex- In case of AWS spot nodes, an executor won't be alive for more than 120 seconds. We can utilize this information in cloud environments and take better decisions about speculating the already running tasks on decommission executors. Yes. This PR adds a new config "spark.executor.decommission.killInterval" which they can explicitly set based on the cloud environment where they are running. Added UT. Closes apache#28619 from prakharjain09/SPARK-21040-speculate-decommission-exec-tasks. Authored-by: Prakhar Jain <[email protected]> Signed-off-by: Holden Karau <[email protected]> [SPARK-20629][CORE][K8S] Copy shuffle data when nodes are being shutdown This pull request adds the ability to migrate shuffle files during Spark's decommissioning. The design document associated with this change is at https://docs.google.com/document/d/1xVO1b6KAwdUhjEJBolVPl9C6sLj7oOveErwDSYdT-pE . To allow this change the `MapOutputTracker` has been extended to allow the location of shuffle files to be updated with `updateMapOutput`. When a shuffle block is put, a block update message will be sent which triggers the `updateMapOutput`. Instead of rejecting remote puts of shuffle blocks `BlockManager` delegates the storage of shuffle blocks to it's shufflemanager's resolver (if supported). A new, experimental, trait is added for shuffle resolvers to indicate they handle remote putting of blocks. The existing block migration code is moved out into a separate file, and a producer/consumer model is introduced for migrating shuffle files from the host as quickly as possible while not overwhelming other executors. Recomputting shuffle blocks can be expensive, we should take advantage of our decommissioning time to migrate these blocks. This PR introduces two new configs parameters, `spark.storage.decommission.shuffleBlocks.enabled` & `spark.storage.decommission.rddBlocks.enabled` that control which blocks should be migrated during storage decommissioning. New unit test & expansion of the Spark on K8s decom test to assert that decommisioning with shuffle block migration means that the results are not recomputed even when the original executor is terminated. This PR is a cleaned-up version of the previous WIP PR I made apache#28331 (thanks to attilapiros for his very helpful reviewing on it :)). Closes apache#28708 from holdenk/SPARK-20629-copy-shuffle-data-when-nodes-are-being-shutdown-cleaned-up. Lead-authored-by: Holden Karau <[email protected]> Co-authored-by: Holden Karau <[email protected]> Co-authored-by: “attilapiros” <[email protected]> Co-authored-by: Attila Zsolt Piros <[email protected]> Signed-off-by: Holden Karau <[email protected]> [SPARK-24266][K8S] Restart the watcher when we receive a version changed from k8s Restart the watcher when it failed with a HTTP_GONE code from the kubernetes api. Which means a resource version has changed. For more relevant information see here: fabric8io/kubernetes-client#1075 No Running spark-submit to a k8s cluster. Not sure how to make an automated test for this. If someone can help me out that would be great. Closes apache#28423 from stijndehaes/bugfix/k8s-submit-resource-version-change. Authored-by: Stijn De Haes <[email protected]> Signed-off-by: Holden Karau <[email protected]> [SPARK-32217] Plumb whether a worker would also be decommissioned along with executor This PR is a giant plumbing PR that plumbs an `ExecutorDecommissionInfo` along with the DecommissionExecutor message. The primary motivation is to know whether a decommissioned executor would also be loosing shuffle files -- and thus it is important to know whether the host would also be decommissioned. In the absence of this PR, the existing code assumes that decommissioning an executor does not loose the whole host with it, and thus does not clear the shuffle state if external shuffle service is enabled. While this may hold in some cases (like K8s decommissioning an executor pod, or YARN container preemption), it does not hold in others like when the cluster is managed by a Standalone Scheduler (Master). This is similar to the existing `workerLost` field in the `ExecutorProcessLost` message. In the future, this `ExecutorDecommissionInfo` can be embellished for knowing how long the executor has to live for scenarios like Cloud spot kills (or Yarn preemption) and the like. No Tweaked an existing unit test in `AppClientSuite` Closes apache#29032 from agrawaldevesh/plumb_decom_info. Authored-by: Devesh Agrawal <[email protected]> Signed-off-by: Holden Karau <[email protected]> [SPARK-32199][SPARK-32198] Reduce job failures during decommissioning This PR reduces the prospect of a job loss during decommissioning. It fixes two holes in the current decommissioning framework: - (a) Loss of decommissioned executors is not treated as a job failure: We know that the decommissioned executor would be dying soon, so its death is clearly not caused by the application. - (b) Shuffle files on the decommissioned host are cleared when the first fetch failure is detected from a decommissioned host: This is a bit tricky in terms of when to clear the shuffle state ? Ideally you want to clear it the millisecond before the shuffle service on the node dies (or the executor dies when there is no external shuffle service) -- too soon and it could lead to some wastage and too late would lead to fetch failures. The approach here is to do this clearing when the very first fetch failure is observed on the decommissioned block manager, without waiting for other blocks to also signal a failure. Without them decommissioning a lot of executors at a time leads to job failures. The task scheduler tracks the executors that were decommissioned along with their `ExecutorDecommissionInfo`. This information is used by: (a) For handling a `ExecutorProcessLost` error, or (b) by the `DAGScheduler` when handling a fetch failure. No Added a new unit test `DecommissionWorkerSuite` to test the new behavior by exercising the Master-Worker decommissioning. I chose to add a new test since the setup logic was quite different from the existing `WorkerDecommissionSuite`. I am open to changing the name of the newly added test suite :-) - Should I add a feature flag to guard these two behaviors ? They seem safe to me that they should only get triggered by decommissioning, but you never know :-). Closes apache#29014 from agrawaldevesh/decom_harden. Authored-by: Devesh Agrawal <[email protected]> Signed-off-by: Holden Karau <[email protected]> [SPARK-32417] Fix flakyness of BlockManagerDecommissionIntegrationSuite This test tries to fix the flakyness of BlockManagerDecommissionIntegrationSuite. Make the block manager decommissioning test be less flaky An interesting failure happens when migrateDuring = true (and persist or shuffle is true): - We schedule the job with tasks on executors 0, 1, 2. - We wait 300 ms and decommission executor 0. - If the task is not yet done on executor 0, it will now fail because the block manager won't be able to save the block. This condition is easy to trigger on a loaded machine where the github checks run. - The task with retry on a different executor (1 or 2) and its shuffle blocks will land there. - No actual block migration happens here because the decommissioned executor technically failed before it could even produce a block. To remove the above race, this change replaces the fixed wait for 300 ms to wait for an actual task to succeed. When a task has succeeded, we know its blocks would have been written for sure and thus its executor would certainly be forced to migrate those blocks when it is decommissioned. The change always decommissions an executor on which a real task finished successfully instead of picking the first executor. Because the system may choose to schedule nothing on the first executor and instead run the two tasks on one executor. I have had bad luck with BlockManagerDecommissionIntegrationSuite and it has failed several times on my PRs. So fixing it. No, unit test only change. Github checks. Ran this test 100 times, 10 at a time in parallel in a script. Closes apache#29226 from agrawaldevesh/block-manager-decom-flaky. Authored-by: Devesh Agrawal <[email protected]> Signed-off-by: Holden Karau <[email protected]> [SPARK-31197][CORE] Shutdown executor once we are done decommissioning Exit the executor when it has been asked to decommission and there is nothing left for it to do. This is a rebase of apache#28817 If we want to use decommissioning in Spark's own scale down we should terminate the executor once finished. Furthermore, in graceful shutdown it makes sense to release resources we no longer need if we've been asked to shutdown by the cluster manager instead of always holding the resources as long as possible. The decommissioned executors will exit and the end of decommissioning. This is sort of a user facing change, however decommissioning hasn't been in any releases yet. I changed the unit test to not send the executor exit message and still wait on the executor exited message. Closes apache#29211 from holdenk/SPARK-31197-exit-execs-redone. Authored-by: Holden Karau <[email protected]> Signed-off-by: Holden Karau <[email protected]> Connect decommissioning to dynamic scaling Because the mock always says there is an RDD we may replicate more than once, and now that there are independent threads Make Spark's dynamic allocation use decommissioning Track the decommissioning executors in the core dynamic scheduler so we don't scale down too low, update the streaming ExecutorAllocationManager to also delegate to decommission Fix up executor add for resource profile Fix our exiting and cleanup thread for better debugging next time. Cleanup the locks we use in decommissioning and clarify some more bits. Verify executors decommissioned, then killed by external external cluster manager are re-launched Verify some additional calls are not occuring in the executor allocation manager suite. Dont' close the watcher until the end of the test Use decommissionExecutors and set adjustTargetNumExecutors to false so that we can match the pattern for killExecutor/killExecutors bump numparts up to 6 Revert "bump numparts up to 6" This reverts commit daf96dd. Small coment & visibility cleanup CR feedback/cleanup Cleanup the merge [SPARK-21040][CORE] Speculate tasks which are running on decommission executors This PR adds functionality to consider the running tasks on decommission executors based on some config. In spark-on-cloud , we sometimes already know that an executor won't be alive for more than fix amount of time. Ex- In AWS Spot nodes, once we get the notification, we know that a node will be gone in 120 seconds. So if the running tasks on the decommissioning executors may run beyond currentTime+120 seconds, then they are candidate for speculation. Currently when an executor is decommission, we stop scheduling new tasks on those executors but the already running tasks keeps on running on them. Based on the cloud, we might know beforehand that an executor won't be alive for more than a preconfigured time. Different cloud providers gives different timeouts before they take away the nodes. For Ex- In case of AWS spot nodes, an executor won't be alive for more than 120 seconds. We can utilize this information in cloud environments and take better decisions about speculating the already running tasks on decommission executors. Yes. This PR adds a new config "spark.executor.decommission.killInterval" which they can explicitly set based on the cloud environment where they are running. Added UT. Closes apache#28619 from prakharjain09/SPARK-21040-speculate-decommission-exec-tasks. Authored-by: Prakhar Jain <[email protected]> Signed-off-by: Holden Karau <[email protected]> [SPARK-20629][CORE][K8S] Copy shuffle data when nodes are being shutdown This pull request adds the ability to migrate shuffle files during Spark's decommissioning. The design document associated with this change is at https://docs.google.com/document/d/1xVO1b6KAwdUhjEJBolVPl9C6sLj7oOveErwDSYdT-pE . To allow this change the `MapOutputTracker` has been extended to allow the location of shuffle files to be updated with `updateMapOutput`. When a shuffle block is put, a block update message will be sent which triggers the `updateMapOutput`. Instead of rejecting remote puts of shuffle blocks `BlockManager` delegates the storage of shuffle blocks to it's shufflemanager's resolver (if supported). A new, experimental, trait is added for shuffle resolvers to indicate they handle remote putting of blocks. The existing block migration code is moved out into a separate file, and a producer/consumer model is introduced for migrating shuffle files from the host as quickly as possible while not overwhelming other executors. Recomputting shuffle blocks can be expensive, we should take advantage of our decommissioning time to migrate these blocks. This PR introduces two new configs parameters, `spark.storage.decommission.shuffleBlocks.enabled` & `spark.storage.decommission.rddBlocks.enabled` that control which blocks should be migrated during storage decommissioning. New unit test & expansion of the Spark on K8s decom test to assert that decommisioning with shuffle block migration means that the results are not recomputed even when the original executor is terminated. This PR is a cleaned-up version of the previous WIP PR I made apache#28331 (thanks to attilapiros for his very helpful reviewing on it :)). Closes apache#28708 from holdenk/SPARK-20629-copy-shuffle-data-when-nodes-are-being-shutdown-cleaned-up. Lead-authored-by: Holden Karau <[email protected]> Co-authored-by: Holden Karau <[email protected]> Co-authored-by: “attilapiros” <[email protected]> Co-authored-by: Attila Zsolt Piros <[email protected]> Signed-off-by: Holden Karau <[email protected]> [SPARK-24266][K8S] Restart the watcher when we receive a version changed from k8s Restart the watcher when it failed with a HTTP_GONE code from the kubernetes api. Which means a resource version has changed. For more relevant information see here: fabric8io/kubernetes-client#1075 No Running spark-submit to a k8s cluster. Not sure how to make an automated test for this. If someone can help me out that would be great. Closes apache#28423 from stijndehaes/bugfix/k8s-submit-resource-version-change. Authored-by: Stijn De Haes <[email protected]> Signed-off-by: Holden Karau <[email protected]> [SPARK-32217] Plumb whether a worker would also be decommissioned along with executor This PR is a giant plumbing PR that plumbs an `ExecutorDecommissionInfo` along with the DecommissionExecutor message. The primary motivation is to know whether a decommissioned executor would also be loosing shuffle files -- and thus it is important to know whether the host would also be decommissioned. In the absence of this PR, the existing code assumes that decommissioning an executor does not loose the whole host with it, and thus does not clear the shuffle state if external shuffle service is enabled. While this may hold in some cases (like K8s decommissioning an executor pod, or YARN container preemption), it does not hold in others like when the cluster is managed by a Standalone Scheduler (Master). This is similar to the existing `workerLost` field in the `ExecutorProcessLost` message. In the future, this `ExecutorDecommissionInfo` can be embellished for knowing how long the executor has to live for scenarios like Cloud spot kills (or Yarn preemption) and the like. No Tweaked an existing unit test in `AppClientSuite` Closes apache#29032 from agrawaldevesh/plumb_decom_info. Authored-by: Devesh Agrawal <[email protected]> Signed-off-by: Holden Karau <[email protected]> [SPARK-32199][SPARK-32198] Reduce job failures during decommissioning This PR reduces the prospect of a job loss during decommissioning. It fixes two holes in the current decommissioning framework: - (a) Loss of decommissioned executors is not treated as a job failure: We know that the decommissioned executor would be dying soon, so its death is clearly not caused by the application. - (b) Shuffle files on the decommissioned host are cleared when the first fetch failure is detected from a decommissioned host: This is a bit tricky in terms of when to clear the shuffle state ? Ideally you want to clear it the millisecond before the shuffle service on the node dies (or the executor dies when there is no external shuffle service) -- too soon and it could lead to some wastage and too late would lead to fetch failures. The approach here is to do this clearing when the very first fetch failure is observed on the decommissioned block manager, without waiting for other blocks to also signal a failure. Without them decommissioning a lot of executors at a time leads to job failures. The task scheduler tracks the executors that were decommissioned along with their `ExecutorDecommissionInfo`. This information is used by: (a) For handling a `ExecutorProcessLost` error, or (b) by the `DAGScheduler` when handling a fetch failure. No Added a new unit test `DecommissionWorkerSuite` to test the new behavior by exercising the Master-Worker decommissioning. I chose to add a new test since the setup logic was quite different from the existing `WorkerDecommissionSuite`. I am open to changing the name of the newly added test suite :-) - Should I add a feature flag to guard these two behaviors ? They seem safe to me that they should only get triggered by decommissioning, but you never know :-). Closes apache#29014 from agrawaldevesh/decom_harden. Authored-by: Devesh Agrawal <[email protected]> Signed-off-by: Holden Karau <[email protected]> [SPARK-32417] Fix flakyness of BlockManagerDecommissionIntegrationSuite This test tries to fix the flakyness of BlockManagerDecommissionIntegrationSuite. Make the block manager decommissioning test be less flaky An interesting failure happens when migrateDuring = true (and persist or shuffle is true): - We schedule the job with tasks on executors 0, 1, 2. - We wait 300 ms and decommission executor 0. - If the task is not yet done on executor 0, it will now fail because the block manager won't be able to save the block. This condition is easy to trigger on a loaded machine where the github checks run. - The task with retry on a different executor (1 or 2) and its shuffle blocks will land there. - No actual block migration happens here because the decommissioned executor technically failed before it could even produce a block. To remove the above race, this change replaces the fixed wait for 300 ms to wait for an actual task to succeed. When a task has succeeded, we know its blocks would have been written for sure and thus its executor would certainly be forced to migrate those blocks when it is decommissioned. The change always decommissions an executor on which a real task finished successfully instead of picking the first executor. Because the system may choose to schedule nothing on the first executor and instead run the two tasks on one executor. I have had bad luck with BlockManagerDecommissionIntegrationSuite and it has failed several times on my PRs. So fixing it. No, unit test only change. Github checks. Ran this test 100 times, 10 at a time in parallel in a script. Closes apache#29226 from agrawaldevesh/block-manager-decom-flaky. Authored-by: Devesh Agrawal <[email protected]> Signed-off-by: Holden Karau <[email protected]> [SPARK-31197][CORE] Shutdown executor once we are done decommissioning Exit the executor when it has been asked to decommission and there is nothing left for it to do. This is a rebase of apache#28817 If we want to use decommissioning in Spark's own scale down we should terminate the executor once finished. Furthermore, in graceful shutdown it makes sense to release resources we no longer need if we've been asked to shutdown by the cluster manager instead of always holding the resources as long as possible. The decommissioned executors will exit and the end of decommissioning. This is sort of a user facing change, however decommissioning hasn't been in any releases yet. I changed the unit test to not send the executor exit message and still wait on the executor exited message. Closes apache#29211 from holdenk/SPARK-31197-exit-execs-redone. Authored-by: Holden Karau <[email protected]> Signed-off-by: Holden Karau <[email protected]> Connect decommissioning to dynamic scaling Because the mock always says there is an RDD we may replicate more than once, and now that there are independent threads Make Spark's dynamic allocation use decommissioning Track the decommissioning executors in the core dynamic scheduler so we don't scale down too low, update the streaming ExecutorAllocationManager to also delegate to decommission Fix up executor add for resource profile Fix our exiting and cleanup thread for better debugging next time. Cleanup the locks we use in decommissioning and clarify some more bits. Verify executors decommissioned, then killed by external external cluster manager are re-launched Verify some additional calls are not occuring in the executor allocation manager suite. Dont' close the watcher until the end of the test Use decommissionExecutors and set adjustTargetNumExecutors to false so that we can match the pattern for killExecutor/killExecutors bump numparts up to 6 Revert "bump numparts up to 6" This reverts commit daf96dd. Small coment & visibility cleanup CR feedback/cleanup Fix up the merge CR feedback, move adjustExecutors to a common utility function Exclude some non-public APIs Remove junk More CR feedback Fix adjustExecutors backport This test fails for me locally and from what I recall it's because we use a different method of resolving the bind address than upstream so disabling the test This test fails for me locally and from what I recall it's because we use a different method of resolving the bind address than upstream so disabling the test Cleanup and drop watcher changes from the backport
jkleckner
pushed a commit
to jkleckner/spark
that referenced
this issue
Nov 3, 2020
…ged from k8s Restart the watcher when it failed with a HTTP_GONE code from the kubernetes api. Which means a resource version has changed. For more relevant information see here: fabric8io/kubernetes-client#1075 No Running spark-submit to a k8s cluster. Not sure how to make an automated test for this. If someone can help me out that would be great. Closes apache#28423 from stijndehaes/bugfix/k8s-submit-resource-version-change. Address review comment to fully qualify import scala.util.control Rebase on branch-3.0 to fix SparkR integration test.
dongjoon-hyun
pushed a commit
to apache/spark
that referenced
this issue
Nov 3, 2020
… changed from k8s ### What changes were proposed in this pull request? This is a straight application of #28423 onto branch-3.0 Restart the watcher when it failed with a HTTP_GONE code from the kubernetes api. Which means a resource version has changed. For more relevant information see here: fabric8io/kubernetes-client#1075 ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? This was tested in #28423 by running spark-submit to a k8s cluster. Closes #29533 from jkleckner/backport-SPARK-24266-to-branch-3.0. Authored-by: Stijn De Haes <[email protected]> Signed-off-by: Dongjoon Hyun <[email protected]>
|
@manusa @rohanKanojia has the Watcher been fixed for the issue? We are using watcher to watch a Kube Job and started noticing this issue. Going through this thread looks like SharedInformer is an alternate. Can a SharedInformer be used to watch a Kube Job? |
|
@SunithaR : Are you talking about try (KubernetesClient client = new DefaultKubernetesClient()) {
// Get Informer Factory
SharedInformerFactory sharedInformerFactory = client.informers();
// Create instance for Job Informer
SharedIndexInformer<Job> jobSharedIndexInformer = sharedInformerFactory.sharedIndexInformerFor(Job.class, JobList.class,
5 * 1000L);
logger.info("Informer factory initialized.");
// Add Event Handler for actions on all Job events received
jobSharedIndexInformer.addEventHandler(
new ResourceEventHandler<Job>() {
@Override
public void onAdd(Job job) {
logger.info("Job " + job.getMetadata().getName() + " got added");
}
@Override
public void onUpdate(Job oldJob, Job newJob) {
logger.info("Job " + oldJob.getMetadata().getName() + " got updated");
}
@Override
public void onDelete(Job job, boolean deletedFinalStateUnknown) {
logger.info("Job " + job.getMetadata().getName() + " got deleted");
}
}
);
logger.info("Starting all registered informers");
sharedInformerFactory.startAllRegisteredInformers();
// Wait for 1 minute
Thread.sleep(60 * 1000L);
logger.info("Stopping informers now..");
sharedInformerFactory.stopAllRegisteredInformers();
} |
|
Thanks @rohanKanojia yes batch/v1 Job. Is there a way to watch a single job? Currently we have this |
|
I think you should be able to do it with something like this: SharedIndexInformer<Job> jobSharedIndexInformer = sharedInformerFactory.sharedIndexInformerFor(Job.class, JobList.class,
new OperationContext().withNamespace(NAMESPACE).withFields(Collections.singletonMap("metadata.name", jobName)),
RESYNC_PERIOD); |
|
Thanks @rohanKanojia will try it out. What is the RESYNC_PERIOD? |
|
Informers maintain their own internal cache. A resync plays back all the events held in the informer's internal cache. So after every resync period informer re-queries API server(list+watch) and updates cache. You can have a look at this blog[0] to see how informer is used with resync period. Recently we also added support for avoiding resync when resync period is set to 0. [0] https://rohaan.medium.com/introduction-to-fabric8-kubernetes-java-client-informer-api-b945082d69af |
|
@manusa @rohanKanojia when sharedInformerFactory.stopAllRegisteredInformers() is called, it produces logs with ERROR severity(see stack trace below). Since we are invoking a graceful shutdown, would be good if these messages are logged at a WARN level, instead of ERROR(signalling major issue) There is a similar issue which has been fixed - kubernetes-client/java#656, but the log level has not been reduced. 2021-01-26 00:21:02.192 ERROR 1 --- [ol-202-thread-1] i.f.k.c.i.cache.ProcessorListener : Processor thread interrupted: null |
|
@SunithaR : Hi, Sorry for late reply. Yes, you're right we should put these error messages as DEBUG/WARN rather than ERROR. Would appreciate if you could create a separate issue for this. We will try to fix this in upcoming sprints. I think it would be awesome if you could contribute a PR for fixing this as it doesn't seem that involved :-) . |
|
@rohanKanojia you got it :), opened new issue - #2753. Will switch log to WARN severity and create PR. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
I am running spark on kubernetes. This is the full issue description https://issues.apache.org/jira/browse/SPARK-24266
I think the exception
too old resource version: 21648111 (21653211)should be better handled in kubernetes-client instead of simply throw it to the caller because resource version is cached by kubernetes-client, not by the caller.kubernetes-client/kubernetes-client/src/main/java/io/fabric8/kubernetes/client/dsl/internal/WatchConnectionManager.java
Lines 259 to 266 in 5b1a57b
The text was updated successfully, but these errors were encountered: