Thank you for the reply. I have added the setting in Test and will update Prod tomorrow then in a time span of two days. we will know the throughput.
This worked and the issue has been fixed now.
Only I am looking for an option to set an alert for 365 days for this configuration.
Started another thread already.
Allright, we will come to know in 2-3 days. We already have restarted the HiveServer2, hence that has renewed the tickets/toekn. So we will know on 6th of April as soon as our users start completing 2-3 days of token creation.
Had the settings go to HiveServer2 only or Metastore safety valve as well?
We are getting the issues again.
Looks like this thread is old, but I would like to add that this was a HiveMetaStore issue, so it should be going into HiveMetaStore safety valve.
@cconnerI'm sorry for digging up this issue after that long time, but we hit exactly the same problem. Following your suggestion we tried set delegation tokens for HS2 and HM, as well as for the Hive service entirely, but in the logs we found:
WARN org.apache.hadoop.hive.conf.HiveConf: [main]: HiveConf of name hive.cluster.delegation.key.update-interval does not exist
WARN org.apache.hadoop.hive.conf.HiveConf: [main]: HiveConf of name hive.cluster.delegation.token.max-lifetime does not exist
WARN org.apache.hadoop.hive.conf.HiveConf: [main]: HiveConf of name hive.cluster.delegation.token.renew-interval does not exist
We are using CDH 5.9.3. Is this parameter still supported in this version? I see these constants exist in the code with default codes. However, searching for solution in the Internet I found that there is another parameter hive.conf.validation that checks if 'hive.' variables are accepted: https://cwiki.apache.org/confluence/display/Hive/Configuration+Properties
Default Value: true
Added In: Hive 0.10.0 with HIVE-2848
Enables type checking for registered Hive configurations. As of Hive 0.14.0 (HIVE-7211), a configuration name that starts with "hive." is regarded as a Hive system property. With hive.conf.validation true (default), any attempts to set a configuration property that starts with "hive." which is not registered to the Hive system will throw an exception.
Is it our case? Or maybe warnings we see in the logs are misleading and actually hive.cluster.delegation.key.* are updated? But we still see in HMS logs messages about new tokens being created/cancelled. I'll be thankful for any clarifications.