You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@flink.apache.org by Anton <an...@yandex.ru> on 2021/11/09 09:22:01 UTC

hbase:meta table usage by flink job

Hello,

 

I'm using HBase to enrich data flow in RichMapFunction and after enabling
Kerberos authorization it turned out that Flink job for some reason needs
access to hbase:meta table. Without access given it was returning errors
like:

INFO  org.apache.hadoop.hbase.client.RpcRetryingCallerImpl         [] - Call
exception, tries=6, retries=16, started=4258 ms ago, cancelled=false,
msg=Call to address=rsb-aspmosddp01.rosbank.rus.socgen/10.45.8.12:16020
failed on local exception: java.io.IOException: java.lang.RuntimeException:
Found no valid authentication method from options, details=row
'table_name,key_value,99999999999999' on table 'hbase:meta' at
region=hbase:meta,,1.1588230740, hostname=dimain_name,16020,1636305734689,
seqNum=-1, see https://s.apache.org/timeout

And therefore job was unable to process any message. Only after giving the
access to hbase:meta the job started to work. But it seems strange because
when I was executing same job from local Intellij IDEA (with same kerberized
hbase and with same keytab file) it worked well even without access to
hbase:meta.

Now I need to justify my access request to hbase:meta on production and I
need to know how it is used by flink and why it is used only when job
executed on server. I couldn't find usage of hbase:meta in Flink source and
HBase documentation says that need of direct access to that table is
'hopefully unlikely'.