[
https://issues.apache.org/jira/browse/FLINK-16267?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17264790#comment-17264790
]
Xintong Song commented on FLINK-16267:
--------------------------------------
Hi [~YordanPavlov],
This ticket is closed because Flink by design does not promise its overall
memory usage stay under the configured `taskmanager.memory.process.size`. As I
explained
[here|https://issues.apache.org/jira/browse/FLINK-16267?focusedCommentId=17109802&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#comment-17109802],
it is not feasible for Flink to put a hard limit on its memory usage. To avoid
such exceeding, you need to properly configure the memory sizes so that Flink
requires a pod that is large enough for your workload.
Concerning how RocksDB use managed memory, one thing to notice is that, RocksDB
does not provide knobs for fully control its memory neither. Flink uses an
estimation based approach to tune the RocksDB parameters and try to get its
memory usage under the expected limit. That means the memory usage might not be
exactly the size reserved (in your case 1833749733 bytes), but should be under
this value in most cases. That is probably why you see the actual memory usage
is less than the reserved size. cc [~yunta], the expert in RocksDBStateBackend,
who may give more detais.
The numbers you provided does not show any RocksDB memory excess. There's one
RocksDB instance per slot, which means as long as the memory usage summed per
TaskManager is below the total managed memory (in your case 3.42GB) there
should be no problem. There could be other implicit memory consumptions that
eventually causes the excess.
Which Flink version are you using? Have you tried upgrading to the lates 1.12
release and see if the problem still exist?
> Flink uses more memory than taskmanager.memory.process.size in Kubernetes
> -------------------------------------------------------------------------
>
> Key: FLINK-16267
> URL: https://issues.apache.org/jira/browse/FLINK-16267
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Task
> Affects Versions: 1.10.0
> Reporter: ChangZhuo Chen (陳昌倬)
> Priority: Major
> Attachments: flink-conf_1.10.0.yaml, flink-conf_1.9.1.yaml,
> oomkilled_taskmanager.log
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> This issue is from
> [https://stackoverflow.com/questions/60336764/flink-uses-more-memory-than-taskmanager-memory-process-size-in-kubernetes]
> h1. Description
> * In Flink 1.10.0, we try to use `taskmanager.memory.process.size` to limit
> the resource used by taskmanager to ensure they are not killed by Kubernetes.
> However, we still get lots of taskmanager `OOMKilled`. The setup is in the
> following section.
> * The taskmanager log is in attachment [^oomkilled_taskmanager.log].
> h2. Kubernete
> * The Kubernetes setup is the same as described in
> [https://ci.apache.org/projects/flink/flink-docs-release-1.10/ops/deployment/kubernetes.html].
> * The following is resource configuration for taskmanager deployment in
> Kubernetes:
> {{resources:}}
> {{ requests:}}
> {{ cpu: 1000m}}
> {{ memory: 4096Mi}}
> {{ limits:}}
> {{ cpu: 1000m}}
> {{ memory: 4096Mi}}
> h2. Flink Docker
> * The Flink docker is built by the following Docker file.
> {{FROM flink:1.10-scala_2.11}}
> RUN mkdir -p /opt/flink/plugins/s3 &&
> ln -s /opt/flink/opt/flink-s3-fs-presto-1.10.0.jar /opt/flink/plugins/s3/
> {{RUN ln -s /opt/flink/opt/flink-metrics-prometheus-1.10.0.jar
> /opt/flink/lib/}}
> h2. Flink Configuration
> * The following are all memory related configurations in `flink-conf.yaml`
> in 1.10.0:
> {{jobmanager.heap.size: 820m}}
> {{taskmanager.memory.jvm-metaspace.size: 128m}}
> {{taskmanager.memory.process.size: 4096m}}
> * We use RocksDB and we don't set `state.backend.rocksdb.memory.managed` in
> `flink-conf.yaml`.
> ** Use S3 as checkpoint storage.
> * The code uses DateStream API
> ** input/output are both Kafka.
> h2. Project Dependencies
> * The following is our dependencies.
> {{val flinkVersion = "1.10.0"}}{{libraryDependencies +=
> "com.squareup.okhttp3" % "okhttp" % "4.2.2"}}
> {{libraryDependencies += "com.typesafe" % "config" % "1.4.0"}}
> {{libraryDependencies += "joda-time" % "joda-time" % "2.10.5"}}
> {{libraryDependencies += "org.apache.flink" %% "flink-connector-kafka" %
> flinkVersion}}
> {{libraryDependencies += "org.apache.flink" % "flink-metrics-dropwizard" %
> flinkVersion}}
> {{libraryDependencies += "org.apache.flink" %% "flink-scala" % flinkVersion
> % "provided"}}
> {{libraryDependencies += "org.apache.flink" %% "flink-statebackend-rocksdb"
> % flinkVersion % "provided"}}
> {{libraryDependencies += "org.apache.flink" %% "flink-streaming-scala" %
> flinkVersion % "provided"}}
> {{libraryDependencies += "org.json4s" %% "json4s-jackson" % "3.6.7"}}
> {{libraryDependencies += "org.log4s" %% "log4s" % "1.8.2"}}
> {{libraryDependencies += "org.rogach" %% "scallop" % "3.3.1"}}
> h2. Previous Flink 1.9.1 Configuration
> * The configuration we used in Flink 1.9.1 are the following. It does not
> have `OOMKilled`.
> h3. Kubernetes
> {{resources:}}
> {{ requests:}}
> {{ cpu: 1200m}}
> {{ memory: 2G}}
> {{ limits:}}
> {{ cpu: 1500m}}
> {{ memory: 2G}}
> h3. Flink 1.9.1
> {{jobmanager.heap.size: 820m}}
> {{taskmanager.heap.size: 1024m}}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
