This is an automated email from the ASF dual-hosted git repository.
wangyang0918 pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/flink.git
The following commit(s) were added to refs/heads/master by this push:
new f503fef [FLINK-25053][docs] Document how to use the usrlib to load
code in the user code class loader
f503fef is described below
commit f503fef9e923453a0eee42f0a955897c06462df9
Author: Lijie Wang <[email protected]>
AuthorDate: Sun Jan 23 20:05:56 2022 +0800
[FLINK-25053][docs] Document how to use the usrlib to load code in the user
code class loader
This closes #18450.
---
.../resource-providers/native_kubernetes.md | 8 +++++
.../resource-providers/standalone/overview.md | 7 +++++
.../docs/deployment/resource-providers/yarn.md | 11 ++++++-
.../docs/ops/debugging/debugging_classloading.md | 34 ++++++++++-----------
.../resource-providers/native_kubernetes.md | 8 +++++
.../resource-providers/standalone/overview.md | 7 +++++
.../docs/deployment/resource-providers/yarn.md | 13 ++++++--
.../docs/ops/debugging/debugging_classloading.md | 35 +++++++++++-----------
8 files changed, 85 insertions(+), 38 deletions(-)
diff --git
a/docs/content.zh/docs/deployment/resource-providers/native_kubernetes.md
b/docs/content.zh/docs/deployment/resource-providers/native_kubernetes.md
index 2d950a5..7ee3302 100644
--- a/docs/content.zh/docs/deployment/resource-providers/native_kubernetes.md
+++ b/docs/content.zh/docs/deployment/resource-providers/native_kubernetes.md
@@ -572,4 +572,12 @@ spec:
emptyDir: { }
```
+### User jars & Classpath
+
+When deploying Flink natively on Kubernetes, the following jars will be
recognized as user-jars and included into user classpath:
+- Session Mode: The JAR file specified in startup command.
+- Application Mode: The JAR file specified in startup command and all JAR
files in Flink's `usrlib` folder.
+
+Please refer to the [Debugging Classloading Docs]({{< ref
"docs/ops/debugging/debugging_classloading"
>}}#overview-of-classloading-in-flink) for details.
+
{{< top >}}
diff --git
a/docs/content.zh/docs/deployment/resource-providers/standalone/overview.md
b/docs/content.zh/docs/deployment/resource-providers/standalone/overview.md
index d97d64f..6498b2c 100644
--- a/docs/content.zh/docs/deployment/resource-providers/standalone/overview.md
+++ b/docs/content.zh/docs/deployment/resource-providers/standalone/overview.md
@@ -236,6 +236,13 @@ Stopping standalonesession daemon (pid: 7349) on host
localhost.
$ bin/stop-zookeeper-quorum.sh
Stopping zookeeper daemon (pid: 7101) on host localhost.</pre>
+### User jars & Classpath
+
+In Standalone mode, the following jars will be recognized as user-jars and
included into user classpath:
+- Session Mode: The JAR file specified in startup command.
+- Application Mode: The JAR file specified in startup command and all JAR
files in Flink's `usrlib` folder.
+
+Please refer to the [Debugging Classloading Docs]({{< ref
"docs/ops/debugging/debugging_classloading"
>}}#overview-of-classloading-in-flink) for details.
{{< top >}}
diff --git a/docs/content.zh/docs/deployment/resource-providers/yarn.md
b/docs/content.zh/docs/deployment/resource-providers/yarn.md
index bb4acbd..7ac23f5 100644
--- a/docs/content.zh/docs/deployment/resource-providers/yarn.md
+++ b/docs/content.zh/docs/deployment/resource-providers/yarn.md
@@ -237,7 +237,14 @@ The configuration parameter for specifying the REST
endpoint port is [rest.bind-
### User jars & Classpath
-By default Flink will include the user jars into the system classpath when
running a single job. This behavior can be controlled with the
[yarn.per-job-cluster.include-user-jar]({{< ref "docs/deployment/config"
>}}#yarn-per-job-cluster-include-user-jar) parameter.
+**Session Mode**
+
+When deploying Flink with Session Mode on Yarn, only the JAR file specified in
startup command will be recognized as user-jars and included into user
classpath.
+
+**PerJob Mode & Application Mode**
+
+When deploying Flink with PerJob/Application Mode on Yarn, the JAR file
specified in startup command and all JAR files in Flink's `usrlib` folder will
be recognized as user-jars.
+By default Flink will include the user-jars into the system classpath. This
behavior can be controlled with the [yarn.classpath.include-user-jar]({{< ref
"docs/deployment/config" >}}#yarn-classpath-include-user-jar) parameter.
When setting this to `DISABLED` Flink will include the jar in the user
classpath instead.
@@ -247,4 +254,6 @@ The user-jars position in the classpath can be controlled
by setting the paramet
- `FIRST`: Adds the jar to the beginning of the system classpath.
- `LAST`: Adds the jar to the end of the system classpath.
+Please refer to the [Debugging Classloading Docs]({{< ref
"docs/ops/debugging/debugging_classloading"
>}}#overview-of-classloading-in-flink) for details.
+
{{< top >}}
diff --git a/docs/content.zh/docs/ops/debugging/debugging_classloading.md
b/docs/content.zh/docs/ops/debugging/debugging_classloading.md
index e2c0512..e08c9c0 100644
--- a/docs/content.zh/docs/ops/debugging/debugging_classloading.md
+++ b/docs/content.zh/docs/ops/debugging/debugging_classloading.md
@@ -33,12 +33,12 @@ When running Flink applications, the JVM will load various
classes over time.
These classes can be divided into three groups based on their origin:
- The **Java Classpath**: This is Java's common classpath, and it includes
the JDK libraries, and all code
- in Flink's `/lib` folder (the classes of Apache Flink and some
dependencies).
+ in Flink's `/lib` folder (the classes of Apache Flink and some
dependencies). They are loaded by *AppClassLoader*.
- The **Flink Plugin Components**: The plugins code in folders under Flink's
`/plugins` folder. Flink's plugin mechanism will dynamically load them once
during startup.
- The **Dynamic User Code**: These are all classes that are included in the
JAR files of dynamically submitted jobs,
- (via REST, CLI, web UI). They are loaded (and unloaded) dynamically per
job.
+ (via REST, CLI, web UI). They are loaded (and unloaded) dynamically by
*FlinkUserCodeClassLoader* per job.
As a general rule, whenever you start the Flink processes first and submit
jobs later, the job's classes are loaded dynamically.
If the Flink processes are started together with the job/application, or if
the application spawns the Flink components (JobManager, TaskManager, etc.),
then all job's classes are in the Java classpath.
@@ -47,10 +47,10 @@ Code in plugin components is loaded dynamically once by a
dedicated class loader
In the following are some more details about the different deployment modes:
-**Standalone Session**
+**Session Mode (Standalone/Yarn/Kubernetes)**
-When starting a Flink cluster as a standalone session, the JobManagers and
TaskManagers are started with the Flink framework classes in the
-Java classpath. The classes from all jobs/applications that are submitted
against the session (via REST / CLI) are loaded *dynamically*.
+When starting a Flink session(Standalone/Yarn/Kubernetes) cluster, the
JobManagers and TaskManagers are started with the Flink framework classes in the
+Java classpath. The classes from all jobs/applications that are submitted
against the session (via REST / CLI) are loaded dynamically by
*FlinkUserCodeClassLoader*.
<!--
**Docker Containers with Flink-as-a-Library**
@@ -61,22 +61,22 @@ created for an job/application and will contain the
job/application's jar files.
-->
-**Docker / Kubernetes Sessions**
+**Per-Job Mode (deprecated) (Yarn)**
-Docker / Kubernetes setups that start first a set of JobManagers /
TaskManagers and then submit jobs/applications via REST or the CLI
-behave like standalone sessions: Flink's code is in the Java classpath, plugin
components are loaded dynamically at startup and the job's code is loaded
dynamically.
+Currently, only Yarn supports Per-Job mode. By default, running a Flink
cluster in Per-Job mode will include the user jars
+(the JAR file specified in startup command and all JAR files in Flink's
`usrlib` folder) into the system classpath (the *AppClassLoader*).
+This behavior can be controlled with the [yarn.classpath.include-user-jar]({{<
ref "docs/deployment/config" >}}#yarn-classpath-include-user-jar) config
option.
+When setting it to `DISABLED`, Flink will include the user jars in the user
classpath and load them dynamically by *FlinkUserCodeClassLoader*.
+See [Flink on Yarn]({{< ref "docs/deployment/resource-providers/yarn" >}}) for
more details.
+**Application Mode (Standalone/Yarn/Kubernetes)**
-**YARN**
+When run a Standalone/Kubernetes Flink cluster in Application Mode, the user
jars (the JAR file specified in startup command and all JAR files in Flink's
`usrlib` folder)
+will be loaded dynamically by *FlinkUserCodeClassLoader*.
-YARN classloading differs between single job deployments and sessions:
-
- - When submitting a Flink job/application directly to YARN (via `bin/flink
run -m yarn-cluster ...`), dedicated TaskManagers and
- JobManagers are started for that job. Those JVMs have user code classes in
the Java classpath.
- That means that there is *no dynamic classloading* involved in that case
for the job.
-
- - When starting a YARN session, the JobManagers and TaskManagers are started
with the Flink framework classes in the
- classpath. The classes from all jobs that are submitted against the
session are loaded dynamically.
+When run a Yarn Flink cluster in Application Mode, the user jars (the JAR file
specified in startup command and all JAR files in Flink's `usrlib` folder)
+will be included into the system classpath (the *AppClassLoader*) by default.
Same as Per-Job mode, when setting the [yarn.classpath.include-user-jar]({{<
ref "docs/deployment/config" >}}#yarn-classpath-include-user-jar)
+to `DISABLED`, Flink will include the user jars in the user classpath and load
them dynamically by *FlinkUserCodeClassLoader*.
## Inverted Class Loading and ClassLoader Resolution Order
diff --git
a/docs/content/docs/deployment/resource-providers/native_kubernetes.md
b/docs/content/docs/deployment/resource-providers/native_kubernetes.md
index a2c3ab0..607f904 100644
--- a/docs/content/docs/deployment/resource-providers/native_kubernetes.md
+++ b/docs/content/docs/deployment/resource-providers/native_kubernetes.md
@@ -576,4 +576,12 @@ spec:
emptyDir: { }
```
+### User jars & Classpath
+
+When deploying Flink natively on Kubernetes, the following jars will be
recognized as user-jars and included into user classpath:
+- Session Mode: The JAR file specified in startup command.
+- Application Mode: The JAR file specified in startup command and all JAR
files in Flink's `usrlib` folder.
+
+Please refer to the [Debugging Classloading Docs]({{< ref
"docs/ops/debugging/debugging_classloading"
>}}#overview-of-classloading-in-flink) for details.
+
{{< top >}}
diff --git
a/docs/content/docs/deployment/resource-providers/standalone/overview.md
b/docs/content/docs/deployment/resource-providers/standalone/overview.md
index 756a4a9..f31ecb4 100644
--- a/docs/content/docs/deployment/resource-providers/standalone/overview.md
+++ b/docs/content/docs/deployment/resource-providers/standalone/overview.md
@@ -285,5 +285,12 @@ $ ./bin/stop-zookeeper-quorum.sh
Stopping zookeeper daemon (pid: 7101) on host localhost.
```
+### User jars & Classpath
+
+In Standalone mode, the following jars will be recognized as user-jars and
included into user classpath:
+- Session Mode: The JAR file specified in startup command.
+- Application Mode: The JAR file specified in startup command and all JAR
files in Flink's `usrlib` folder.
+
+Please refer to the [Debugging Classloading Docs]({{< ref
"docs/ops/debugging/debugging_classloading"
>}}#overview-of-classloading-in-flink) for details.
{{< top >}}
diff --git a/docs/content/docs/deployment/resource-providers/yarn.md
b/docs/content/docs/deployment/resource-providers/yarn.md
index 0a803bb..a967cb3 100644
--- a/docs/content/docs/deployment/resource-providers/yarn.md
+++ b/docs/content/docs/deployment/resource-providers/yarn.md
@@ -1,4 +1,4 @@
----
+\---
title: YARN
weight: 5
type: docs
@@ -253,7 +253,14 @@ The configuration parameter for specifying the REST
endpoint port is [rest.bind-
### User jars & Classpath
-By default Flink will include the user jars into the system classpath when
running a single job. This behavior can be controlled with the
[yarn.per-job-cluster.include-user-jar]({{< ref "docs/deployment/config"
>}}#yarn-per-job-cluster-include-user-jar) parameter.
+**Session Mode**
+
+When deploying Flink with Session Mode on Yarn, only the JAR file specified in
startup command will be recognized as user-jars and included into user
classpath.
+
+**PerJob Mode & Application Mode**
+
+When deploying Flink with PerJob/Application Mode on Yarn, the JAR file
specified in startup command and all JAR files in Flink's `usrlib` folder will
be recognized as user-jars.
+By default Flink will include the user-jars into the system classpath. This
behavior can be controlled with the [yarn.classpath.include-user-jar]({{< ref
"docs/deployment/config" >}}#yarn-classpath-include-user-jar) parameter.
When setting this to `DISABLED` Flink will include the jar in the user
classpath instead.
@@ -263,4 +270,6 @@ The user-jars position in the classpath can be controlled
by setting the paramet
- `FIRST`: Adds the jar to the beginning of the system classpath.
- `LAST`: Adds the jar to the end of the system classpath.
+Please refer to the [Debugging Classloading Docs]({{< ref
"docs/ops/debugging/debugging_classloading"
>}}#overview-of-classloading-in-flink) for details.
+
{{< top >}}
diff --git a/docs/content/docs/ops/debugging/debugging_classloading.md
b/docs/content/docs/ops/debugging/debugging_classloading.md
index 0bb26b2..7a65bff 100644
--- a/docs/content/docs/ops/debugging/debugging_classloading.md
+++ b/docs/content/docs/ops/debugging/debugging_classloading.md
@@ -33,12 +33,12 @@ When running Flink applications, the JVM will load various
classes over time.
These classes can be divided into three groups based on their origin:
- The **Java Classpath**: This is Java's common classpath, and it includes
the JDK libraries, and all code
- in Flink's `/lib` folder (the classes of Apache Flink and some
dependencies).
+ in Flink's `/lib` folder (the classes of Apache Flink and some
dependencies). They are loaded by *AppClassLoader*.
- The **Flink Plugin Components**: The plugins code in folders under Flink's
`/plugins` folder. Flink's plugin mechanism will dynamically load them once
during startup.
- The **Dynamic User Code**: These are all classes that are included in the
JAR files of dynamically submitted jobs,
- (via REST, CLI, web UI). They are loaded (and unloaded) dynamically per
job.
+ (via REST, CLI, web UI). They are loaded (and unloaded) dynamically by
*FlinkUserCodeClassLoader* per job.
As a general rule, whenever you start the Flink processes first and submit
jobs later, the job's classes are loaded dynamically.
If the Flink processes are started together with the job/application, or if
the application spawns the Flink components (JobManager, TaskManager, etc.),
then all job's classes are in the Java classpath.
@@ -47,10 +47,10 @@ Code in plugin components is loaded dynamically once by a
dedicated class loader
In the following are some more details about the different deployment modes:
-**Standalone Session**
+**Session Mode (Standalone/Yarn/Kubernetes)**
-When starting a Flink cluster as a standalone session, the JobManagers and
TaskManagers are started with the Flink framework classes in the
-Java classpath. The classes from all jobs/applications that are submitted
against the session (via REST / CLI) are loaded *dynamically*.
+When starting a Flink session(Standalone/Yarn/Kubernetes) cluster, the
JobManagers and TaskManagers are started with the Flink framework classes in the
+Java classpath. The classes from all jobs/applications that are submitted
against the session (via REST / CLI) are loaded dynamically by
*FlinkUserCodeClassLoader*.
<!--
**Docker Containers with Flink-as-a-Library**
@@ -61,23 +61,22 @@ created for an job/application and will contain the
job/application's jar files.
-->
-**Docker / Kubernetes Sessions**
+**Per-Job Mode (deprecated) (Yarn)**
-Docker / Kubernetes setups that start first a set of JobManagers /
TaskManagers and then submit jobs/applications via REST or the CLI
-behave like standalone sessions: Flink's code is in the Java classpath, plugin
components are loaded dynamically at startup and the job's code is loaded
dynamically.
+Currently, only Yarn supports Per-Job mode. By default, running a Flink
cluster in Per-Job mode will include the user jars
+(the JAR file specified in startup command and all JAR files in Flink's
`usrlib` folder) into the system classpath (the *AppClassLoader*).
+This behavior can be controlled with the [yarn.classpath.include-user-jar]({{<
ref "docs/deployment/config" >}}#yarn-classpath-include-user-jar) config
option.
+When setting it to `DISABLED`, Flink will include the user jars in the user
classpath and load them dynamically by *FlinkUserCodeClassLoader*.
+See [Flink on Yarn]({{< ref "docs/deployment/resource-providers/yarn" >}}) for
more details.
+**Application Mode (Standalone/Yarn/Kubernetes)**
-**YARN**
-
-YARN classloading differs between single job deployments and sessions:
-
- - When submitting a Flink job/application directly to YARN (via `bin/flink
run -m yarn-cluster ...`), dedicated TaskManagers and
- JobManagers are started for that job. Those JVMs have user code classes in
the Java classpath.
- That means that there is *no dynamic classloading* involved in that case
for the job.
-
- - When starting a YARN session, the JobManagers and TaskManagers are started
with the Flink framework classes in the
- classpath. The classes from all jobs that are submitted against the
session are loaded dynamically.
+When run a Standalone/Kubernetes Flink cluster in Application Mode, the user
jars (the JAR file specified in startup command and all JAR files in Flink's
`usrlib` folder)
+will be loaded dynamically by *FlinkUserCodeClassLoader*.
+When run a Yarn Flink cluster in Application Mode, the user jars (the JAR file
specified in startup command and all JAR files in Flink's `usrlib` folder)
+will be included into the system classpath (the *AppClassLoader*) by default.
Same as Per-Job mode, when setting the [yarn.classpath.include-user-jar]({{<
ref "docs/deployment/config" >}}#yarn-classpath-include-user-jar)
+to `DISABLED`, Flink will include the user jars in the user classpath and load
them dynamically by *FlinkUserCodeClassLoader*.
## Inverted Class Loading and ClassLoader Resolution Order
