[
https://issues.apache.org/jira/browse/HDDS-9898?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17796740#comment-17796740
]
Mohammad Arafat Khan commented on HDDS-9898:
--------------------------------------------
I tried replicating this locally on a docker setup. I created an EC container
*Container 1* which was initially failing to register but later registering
successfully due to the absence of a required pipeline. Here's a summary:
# Initially, there were multiple warnings and errors indicating that Container
1 couldn't be added due to a missing pipeline
({*}PipelineID=68426b76-65ec-4663-be73-64ac4e1f8412{*}). The system recognized
the need to register Container 1 but couldn't do so because the pipeline was
missing.
# Later in the log, a new pipeline with the required ID
({*}PipelineID=68426b76-65ec-4663-be73-64ac4e1f8412{*}) was created or
recognized by the system.
# After the pipeline was established, Container 1 was successfully added to
the system.
In summary, the issue was resolved when the missing pipeline was created,
allowing Container 1 to register successfully. This might have been caused by a
delay in pipeline creation by the *{{PipeLineSyncTask}}* of R{*}econ{*} for EC
containers in recon.
What we need to investigate now is that why is the delay happening for EC
container pipelines and not the ratis pipelines.
> Recon SCM doesn't store EC containers
> -------------------------------------
>
> Key: HDDS-9898
> URL: https://issues.apache.org/jira/browse/HDDS-9898
> Project: Apache Ozone
> Issue Type: Bug
> Reporter: Christos Bisias
> Assignee: Mohammad Arafat Khan
> Priority: Major
>
> While working on an integration test for
> https://issues.apache.org/jira/browse/HDDS-9645
> I created an EC container and then tried to compare SCM to Recon. Recon-SCM
> wasn't picking up the container and I noticed this ERROR in the logs.
> {code:java}
> 2023-12-11 15:57:38,663 [Recon-FixedThreadPoolWithAffinityExecutor-0-0] ERROR
> container.ContainerReportHandler
> (ContainerReportHandler.java:processSingleReplica(228)) - Received container
> report for an unknown container 3 from datanode
> 7501f622-8ca7-4a47-a65e-e2b3d77723c4(localhost/127.0.0.1).{code}
> I tested it with the docker env on master and Recon doesn't store EC
> containers. When creating EC containers Recon's SCM containers table is empty.
> {code:java}
> bash-4.2$ ozone freon ockg -t 10 -n 10 --type=EC -r=rs-3-2-1024k
> ...
> ...
> ...
> bash-4.2$ ozone debug ldb --db=/data/metadata/recon-scm.db scan
> --cf=containers
> { } {code}
> Recon UI also displays 0 containers. When creating Ratis containers, it works
> {code:java}
> bash-4.2$ ozone freon ockg -t 10 -n 10
> ...
> ...
> ...
> Total execution time (sec): 2
> Failures: 0
> Successful executions: 10
> bash-4.2$ ozone debug ldb --db=/data/metadata/recon-scm.db scan
> --cf=containers
> { {
> "id" : 6
> }: {
> "state" : "OPEN",
> "stateEnterTime" : 1702302641.257000000,
> "replicationConfig" : {
> "replicationFactor" : "ONE",
> "replicationType" : "RATIS"
> },
> "usedBytes" : 0,
> "numberOfKeys" : 0,
> "lastUsed" : 1702302648.723516000,
> "owner" : "omServiceIdDefault",
> "containerID" : 6,
> "deleteTransactionId" : 0,
> "sequenceId" : 0
> }
> , {
> ...{code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
