Xianxun Ye created FLINK-32817:
----------------------------------
Summary: Supports running jar file names with Spaces
Key: FLINK-32817
URL: https://issues.apache.org/jira/browse/FLINK-32817
Project: Flink
Issue Type: Improvement
Components: Deployment / YARN
Affects Versions: 1.14.0
Reporter: Xianxun Ye
When submitting a flink jar to a yarn cluster, if the jar filename has spaces
in it, the task will not be able to successfully parse the file path in
`YarnLocalResourceDescriptor`, and the following exception will occur in
JobManager.
The Flink jar file name is: StreamSQLExample 2.jar
{code:java}
2023-08-09 18:54:31,787 WARN
org.apache.flink.runtime.extension.resourcemanager.NeActiveResourceManager [] -
Failed requesting worker with resource spec WorkerResourceSpec {cpuCores=1.0,
taskHeapSize=220.160mb (230854450 bytes), taskOffHeapSize=0 bytes,
networkMemSize=158.720mb (166429984 bytes), managedMemSize=952.320mb (998579934
bytes), numSlots=1}, current pending count: 0
java.util.concurrent.CompletionException: org.apache.flink.util.FlinkException:
Error to parse YarnLocalResourceDescriptor from
YarnLocalResourceDescriptor{key=StreamSQLExample 2.jar,
path=hdfs://***/.flink/application_1586413220781_33151/StreamSQLExample 2.jar,
size=7937, modificationTime=1691578403748, visibility=APPLICATION, type=FILE}
at
org.apache.flink.util.concurrent.FutureUtils.lambda$supplyAsync$21(FutureUtils.java:1052)
~[flink-dist_2.12-1.14.0.jar:1.14.0]
at
java.util.concurrent.CompletableFuture$AsyncSupply.run(CompletableFuture.java:1590)
~[?:1.8.0_152]
at
java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
~[?:1.8.0_152]
at
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
~[?:1.8.0_152]
at java.lang.Thread.run(Thread.java:748) ~[?:1.8.0_152]
Caused by: org.apache.flink.util.FlinkException: Error to parse
YarnLocalResourceDescriptor from
YarnLocalResourceDescriptor{key=StreamSQLExample 2.jar,
path=hdfs://sloth-jd-pub/user/sloth/.flink/application_1586413220781_33151/StreamSQLExample
2.jar, size=7937, modificationTime=1691578403748, visibility=APPLICATION,
type=FILE}
at
org.apache.flink.yarn.YarnLocalResourceDescriptor.fromString(YarnLocalResourceDescriptor.java:112)
~[flink-dist_2.12-1.14.0.jar:1.14.0]
at
org.apache.flink.yarn.Utils.decodeYarnLocalResourceDescriptorListFromString(Utils.java:600)
~[flink-dist_2.12-1.14.0.jar:1.14.0]
at org.apache.flink.yarn.Utils.createTaskExecutorContext(Utils.java:491)
~[flink-dist_2.12-1.14.0.jar:1.14.0]
at
org.apache.flink.yarn.YarnResourceManagerDriver.createTaskExecutorLaunchContext(YarnResourceManagerDriver.java:452)
~[flink-dist_2.12-1.14.0.jar:1.14.0]
at
org.apache.flink.yarn.YarnResourceManagerDriver.lambda$startTaskExecutorInContainerAsync$1(YarnResourceManagerDriver.java:383)
~[flink-dist_2.12-1.14.0.jar:1.14.0]
at
org.apache.flink.util.concurrent.FutureUtils.lambda$supplyAsync$21(FutureUtils.java:1050)
~[flink-dist_2.12-1.14.0.jar:1.14.0]
... 4 more{code}
>From what I understand, the HDFS cluster allows for file names with spaces, as
>well as S3.
I think we could replace the `LOCAL_RESOURCE_DESC_FORMAT` for
{code:java}
// code placeholder
private static final Pattern LOCAL_RESOURCE_DESC_FORMAT =
Pattern.compile(
"YarnLocalResourceDescriptor\\{"
+ "key=([\\S\\x20]+), path=([\\S\\x20]+),
size=([\\d]+), modificationTime=([\\d]+), visibility=(\\S+), type=(\\S+)}");
{code}
add '\x20' to only match the spaces
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
