wecharyu opened a new pull request, #10886:
URL: https://github.com/apache/hudi/pull/10886
### Change Logs
When `spark.speculation` is enabled, if the write metadata operation become
slow for some reason, a speculative will be started to write the same metadata
file concurrently.
In HDFS, two threads working on same piece may cause the issue:
```bash
Caused by:
org.apache.hadoop.ipc.RemoteException(java.io.FileNotFoundException): File does
not exist: /path/to/table/a=3519/b=3520/c=3521/.hoodie_partition_metadata_112
(inode 48415575374) Holder DFSClient_NONMAPREDUCE_-2108606624_29 does not have
any open files.
at
org.apache.hadoop.hdfs.server.namenode.FSNamesystem.checkLease(FSNamesystem.java:3239)
at
org.apache.hadoop.hdfs.server.namenode.FSDirWriteFileOp.analyzeFileState(FSDirWriteFileOp.java:617)
at
org.apache.hadoop.hdfs.server.namenode.FSDirWriteFileOp.validateAddBlock(FSDirWriteFileOp.java:176)
at
org.apache.hadoop.hdfs.server.namenode.FSNamesystem.getAdditionalBlock(FSNamesystem.java:3116)
at
org.apache.hadoop.hdfs.server.namenode.NameNodeRpcServer.addBlock(NameNodeRpcServer.java:1075)
at
org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolServerSideTranslatorPB.addBlock(ClientNamenodeProtocolServerSideTranslatorPB.java:619)
at
org.apache.hadoop.ipc.ProtobufRpcEngine2$Server$ProtoBufRpcInvoker.call(ProtobufRpcEngine2.java:623)
at
org.apache.hadoop.ipc.ProtobufRpcEngine2$Server$ProtoBufRpcInvoker.call(ProtobufRpcEngine2.java:591)
at
org.apache.hadoop.ipc.ProtobufRpcEngine2$Server$ProtoBufRpcInvoker.call(ProtobufRpcEngine2.java:575)
at org.apache.hadoop.ipc.RPC$Server.call(RPC.java:1137)
at org.apache.hadoop.ipc.Server$RpcCall.run(Server.java:1230)
at org.apache.hadoop.ipc.Server$RpcCall.run(Server.java:1128)
at javax.security.auth.Subject.doAs(Subject.java:422)
at
org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:2005)
at org.apache.hadoop.ipc.Server$Handler.run(Server.java:3354)
```
Then the partition metadata file can not be created successfully.
We try to fix this issue by:
1. Use a unique token in `HoodiePartitionMetadata#trySave()` to support
concurrent write
2. Throw exception when failed to write partition metadata file.
### Impact
Fix the concurrently write same partition metadata issue.
### Risk level (write none, low medium or high below)
Low
### Documentation Update
_Describe any necessary documentation update if there is any new feature,
config, or user-facing change. If not, put "none"._
- _The config description must be updated if new configs are added or the
default value of the configs are changed_
- _Any new feature or user-facing change requires updating the Hudi website.
Please create a Jira ticket, attach the
ticket number here and follow the
[instruction](https://hudi.apache.org/contribute/developer-setup#website) to
make
changes to the website._
### Contributor's checklist
- [ ] Read through [contributor's
guide](https://hudi.apache.org/contribute/how-to-contribute)
- [ ] Change Logs and Impact were stated clearly
- [ ] Adequate tests were added if applicable
- [ ] CI passed
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
