[
https://issues.apache.org/jira/browse/FLINK-6522?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16609205#comment-16609205
]
Ufuk Celebi edited comment on FLINK-6522 at 9/10/18 1:39 PM:
-------------------------------------------------------------
[~till.rohrmann] I think I ran into this issue as well. Can you please confirm
that the following indeed describes the same issue:
1. I create an HA standalone cluster that has a job-specific cluster ID (e.g.
the job ID)
2. I run the job and see job-specific data in the storageDir and ZK
3. I cancel the job and stop the cluster
4. I *only* see data in storageDir be cleaned up, but still have the following
nodes in ZK (with each sub-node such as jobgraphs being empty):
{code:java}
/flink/6ecc3c8b-5189-4c2c-ae61-88c31b8b3d06/[jobgraphs, leader, checkpoints,
leaderlatch, checkpoint-counter, running_job_registry]
{code}
was (Author: uce):
[~till.rohrmann] I think I ran into this issue as well. Can you please confirm
that the following is indeed describes the same issue:
1. I create an HA standalone cluster that has a job-specific cluster ID (e.g.
the job ID)
2. I run the job and see job-specific data in the storageDir and ZK
3. I cancel the job and stop the cluster
4. I *only* see data in storageDir be cleaned up, but still have the following
nodes in ZK (with each sub-node such as jobgraphs being empty):
{code:java}
/flink/6ecc3c8b-5189-4c2c-ae61-88c31b8b3d06/[jobgraphs, leader, checkpoints,
leaderlatch, checkpoint-counter, running_job_registry]
{code}
> Add ZooKeeper cleanup logic to ZooKeeperHaServices
> --------------------------------------------------
>
> Key: FLINK-6522
> URL: https://issues.apache.org/jira/browse/FLINK-6522
> Project: Flink
> Issue Type: Improvement
> Components: Distributed Coordination
> Affects Versions: 1.3.0, 1.4.0
> Reporter: Till Rohrmann
> Assignee: Fang Yong
> Priority: Major
>
> The {{ZooKeeperHaServices}} provide a {{CuratorFramework}} client to access
> ZooKeeper data. Consequently, all data (also for different job) are stored
> under the same root node. When
> {{HighAvailabilityServices#closeAndCleanupAllData}} is called, then this data
> should be cleaned up. This cleanup logic is currently missing.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
