Closed vansimonsen closed 3 years ago
Issue (https://github.com/apache/hudi/issues/2801) might be a duplicate.
However while creating an unpartitioned table, my dataframe.write is getting succeeded but I am not able to query the data via hive. Although spark read are working fine for me though. (Testing via spark shell and I am using jdbc to connect to hive)
@vansimonsen Can you check the issue that @aditiwari01 is pointing to and check if you are using the correct KeyGenerators
as well as PartitionValueExtractor
(check here -> https://hudi.apache.org/docs/configurations.html#HIVE_PARTITION_EXTRACTOR_CLASS_OPT_KEY) ?
Additionally, this looks like the basePath
might not have been correctly registered to Glue. Let me know after you check these configs, if they don't work, this may be a legit bug
its might be related to missing Glue database s3 path, the field is named "Amazon S3 path"(lakeformation) or "Location"(glue) in aws console
as far as i see at one point in code it is tryiong to construct a path like : getDatabasePath
+tableName
in my case it was creating: s3://MyBucketMytable because of missing /
. at the end of the database Location
@ismailsimsek Are you saying it was fixed after you fixed the databasePath / location in your glue metastore to include /
? Is the /
expected always at the end of the path ? If yes, we can probably put in that fix in hudi hive sync.
@vansimonsen Can you check if this is the root cause for you ?
@ismailsimsek @vansimonsen Closing this due to inactivity, please re-open it or open a new one if you need further assistance.
I am facing the same issue. Please can you share the fix. I am using Hudi version 0.8.
Describe the problem you faced
0.6.0
,0.7.0
and0.8.0
)Previous fix is implemented on new versions, but it continues failing
To Reproduce
Steps to reproduce the behavior:
Expected behavior
The table would be created without throw the exception, without any partition or
default
partitionpathEnvironment Description
Hudi version :
0.6.0
,0.7.0
and0.8.0
Spark version :
2.4.7
Hive version : Aws glue data catalog integration on EMR
Hadoop version : Amazon Hadoop distribution
Storage (HDFS/S3/GCS..) : S3
Running on Docker? (yes/no) : no
Stacktrace