Reply
Contributor
Posts: 31
Registered: ‎06-26-2015

ResourceManager dynamic pool update fails due to invalid kerberos principal

When I update dynamic resource pool, CM automatically tries to update it. however it fails with following error

 

Completed only 4/5 steps. First failure: Process yarn-RESOURCEMANAGER-refresh (id=2101) on host host (id=3) exited with 255 and expected 0

 

on stderr:

User yarn/host@REALM (auth:KERBEROS) is not authorized for protocol interface 

 

also, on RM server role log, I find many of this log

2015-07-09 08:37:48,675 WARN org.apache.hadoop.security.authentication.server.AuthenticationFilter: AuthenticationToken ignored: AuthenticationToken expired

 

 

any thoughts?

Contributor
Posts: 31
Registered: ‎06-26-2015

Re: ResourceManager dynamic pool update fails due to invalid kerberos principal

i also found this error

 

using yarn/host@REALM as Kerberos principal
using /var/run/cloudera-scm-agent/process/2109-yarn-RESOURCEMANAGER-refresh/krb5cc_489 as Kerberos ticket cache

 

 

so i went to RM host and ran "klist -c /var/run/cloudera-scm-agent/process/2109-yarn-RESOURCEMANAGER-refresh/krb5cc_489"

 

Ticket cache: FILE:/var/run/cloudera-scm-agent/process/2109-yarn-RESOURCEMANAGER-refresh/krb5cc_489
Default principal: yarn/host@REALM

Valid starting     Expires            Service principal
07/09/15 08:53:24  07/10/15 08:53:24  krbtgt/REALM@REALM
        renew until 07/16/15 08:53:23

 

this ticket seems to be ok

Explorer
Posts: 17
Registered: ‎12-07-2018

Re: ResourceManager dynamic pool update fails due to invalid kerberos principal

Please tell me after the Refresh Dynamic Resource Pools there is an error
"Failed to perform refresh of roles in YARN, Impala, and (in CDH 5.8 and later) Hive services."
Refresh the ResourceManager's pool ACLs, states, scheduler specific properties, and hosts allow/exclude information.
Process yarn-RESOURCEMANAGER-refresh (id=12038) on host bda3node04.vtb24.ru (id=6) exited with 255 and expected 0
Refresh the ResourceManager's pool ACLs, states, scheduler specific properties, and hosts allow/exclude information.
Process yarn-RESOURCEMANAGER-refresh (id=12039) on host bda3node03.vtb24.ru (id=5) exited with 255 and expected 0

How to deal with this?

Announcements