Created on 12-25-2016 07:24 PM
ERRORS:
from HDFS log:016-10-30 17:44:04,226 ERROR impl.CloudSolrClient (CloudSolrClient.java:requestWithRetryOnStaleState(903)) - Request to collection ranger_audits failed due to (401) org.apache.solr.client.solrj.impl.HttpSolrClient$RemoteSolrException: Error from server at https://hwx.com:8886/solr/ranger_audits_shard1_replica1: Expected mime type application/octet-stream but got text/html. <html> <head> <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/> <title>Error 401 Authentication required</title> </head> <body><h2>HTTP ERROR 401</h2> <p>Problem accessing /solr/ranger_audits_shard1_replica1/update. Reason: <pre> Authentication required</pre></p><hr><i><small>Powered by Jetty://</small></i><hr/>From Hive log:
016-10-30 17:50:46,189 WARN [org.apache.ranger.audit.queue.AuditBatchQueue1]: provider.BaseAuditHandler (BaseAuditHandler.java:logFailedEvent(374)) - failed to log audit event: {"repoType":3,"repo":"AA_Prod_hive","reqUser":"alex","evtTime":"2016-10-30 17:50:43.587","access":"USE","resource":"default","resType":"@ database","action":"_any","result":0,"policy":-1,"enforcer":"ranger-acl","sess":"cf4d0c81-c4df-483b-ab51-aa7bb5cb1633","cliType":"HIVESERVER2","cliIP":"172.26 .205.88","reqData":"show tables","agentHost":"hwx.com","logType":"RangerAudit","id":"d41d25ee-d198-475d-a288-11d6cc76535c-0","seq_num":1 ,"event_count":1,"event_dur_ms":0,"tags":[],"additional_info":"{\"remote-ip-address\":172.26.1.1, \"forwarded-ip-addresses\":[]"} org.apache.solr.client.solrj.impl.CloudSolrClient$RouteException: Error from server at https://hwx:8886/solr/ranger_audits_shard1_re plica1: Expected mime type application/octet-stream but got text/html. <html> <head> <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/> <title>Error 401 Authentication required</title>Debug: Once DEBUG log for krb5 is enabled (-Dsun.security.krb5.debug=true) we can see in both hdfs (log is hadoop-hdfs-namenode-hwx.com.out) and hive (log is hive-server2.out) the same issue as from the Knox ticket (it tries to use the HTTPS/_HOST principal instead of HTTP/_HOST as it's standard with spnego):
>>KRBError: sTime is Sun Oct 30 17:50:08 PDT 2016 1477875008000 suSec is 518135 error code is 7 error Message is Server not found in Kerberos database sname is HTTPS/host@HWX.COM msgType is 30ROOT CAUSE: There is a defect in httpclient 4.5.2 that got introduced in HDP 2.5. WORKAROUND: Downgrade all the httpclient at 4.5.2 for ranger to 4.5.1
This will be fixed in a future release.
Created on 12-25-2016 07:28 PM
Created on 12-25-2016 07:30 PM
Created on 03-16-2017 10:39 AM
Hi,
This bug can have consequences on Spark / Yarn as well. We were encountering Out of Memory conditions running Spark job, not matter how much memory we assigned, we kept ending up exhausting it completely.
This behaviour actually disappeared when we applied the fix listed here.
I'll post back when I know more about the root cause & link between issues.
Regards,
Christophe