[
https://issues.apache.org/jira/browse/FLINK-2354?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14941173#comment-14941173
]
ASF GitHub Bot commented on FLINK-2354:
---------------------------------------
Github user tillrohrmann commented on a diff in the pull request:
https://github.com/apache/flink/pull/1153#discussion_r41024851
--- Diff:
flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/CheckpointCoordinator.java
---
@@ -110,61 +120,68 @@
public CheckpointCoordinator(
JobID job,
- int numSuccessfulCheckpointsToRetain,
long checkpointTimeout,
ExecutionVertex[] tasksToTrigger,
ExecutionVertex[] tasksToWaitFor,
ExecutionVertex[] tasksToCommitTo,
- ClassLoader userClassLoader) {
+ ClassLoader userClassLoader,
+ CheckpointIDCounter checkpointIDCounter,
+ CompletedCheckpoints completedCheckpoints,
+ RecoveryMode recoveryMode) throws Exception {
// some sanity checks
if (job == null || tasksToTrigger == null ||
tasksToWaitFor == null || tasksToCommitTo ==
null) {
throw new NullPointerException();
}
- if (numSuccessfulCheckpointsToRetain < 1) {
- throw new IllegalArgumentException("Must retain at
least one successful checkpoint");
- }
if (checkpointTimeout < 1) {
throw new IllegalArgumentException("Checkpoint timeout
must be larger than zero");
}
this.job = job;
--- End diff --
Maybe we could harmonize the null checking as you've done it.
> Recover running jobs on JobManager failure
> ------------------------------------------
>
> Key: FLINK-2354
> URL: https://issues.apache.org/jira/browse/FLINK-2354
> Project: Flink
> Issue Type: Sub-task
> Components: JobManager
> Affects Versions: master
> Reporter: Ufuk Celebi
> Assignee: Ufuk Celebi
> Fix For: 0.10
>
>
> tl;dr Persist JobGraphs in state backend and coordinate reference to state
> handle via ZooKeeper.
> Problem: When running multiple JobManagers in high availability mode, the
> leading job manager looses all running jobs when it fails. After a new
> leading job manager is elected, it is not possible to recover any previously
> running jobs.
> Solution: The leading job manager, which receives the job graph writes 1) the
> job graph to a state backend, and 2) a reference to the respective state
> handle to ZooKeeper. In general, job graphs can become large (multiple MBs,
> because they include closures etc.). ZooKeeper is not designed for data of
> this size. The level of indirection via the reference to the state backend
> keeps the data in ZooKeeper small.
> Proposed ZooKeeper layout:
> /flink (default)
> +- currentJobs
> +- job id i
> +- state handle reference of job graph i
> The 'currentJobs' node needs to be persistent to allow recovery of jobs
> between job managers. The currentJobs node needs to satisfy the following
> invariant: There is a reference to a job graph with id i IFF the respective
> job graph needs to be recovered by a newly elected job manager leader.
> With this in place, jobs will be recovered from their initial state (as if
> resubmitted). The next step is to backup the runtime state handles of
> checkpoints in a similar manner.
> ---
> This work will be based on [~trohrm...@apache.org]'s implementation of
> FLINK-2291. The leader election service notifies the job manager about
> granted/revoked leadership. This notification happens via Akka and thus
> serially *per* job manager, but results in eventually consistent state
> between job managers. For some snapshots of time it is possible to have a new
> leader granted leadership, before the old one has been revoked its leadership.
> [~trohrm...@apache.org], can you confirm that leadership does not guarantee
> mutually exclusive access to the shared 'currentJobs' state?
> For example, the following can happen:
> - JM 1 is leader, JM 2 is standby
> - JOB i is running (and hence /flink/currentJobs/i exists)
> - ZK notifies leader election service (LES) of JM 1 and JM 2
> - LES 2 immediately notifies JM 2 about granted leadership, but LES 1
> notification revoking leadership takes longer
> - JOB i finishes (TMs don't notice leadership change yet) and JM 1 receives
> final JobStatusChange
> - JM 2 resubmits the job /flink/currentJobs/i
> - JM 1 removes /flink/currentJobs/i, because it is now finished
> => inconsistent state (wrt the specified invariant above)
> If it is indeed a problem, we can circumvent this with a Curator recipe for
> [shared locks|http://curator.apache.org/curator-recipes/shared-lock.html] to
> coordinate the access to currentJobs. The lock needs to be acquired on
> leadership.
> ---
> Minimum required tests:
> - Unit tests for job graph serialization and writing to state backend and
> ZooKeeper with expected nodes
> - Unit tests for job submission to job manager in leader/non-leader state
> - Unit tests for leadership granting/revoking and job submission/restarting
> interleavings
> - Process failure integration tests with single and multiple running jobs
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
