Hi Andrew, We use keytab file for hbase, there's no ticket cache file under /tmp, but we still encounter the same error as above, how about this?
2014-03-12 7:37 GMT+08:00 Andrew Purtell <[email protected]>: > If you might have more than one security enabled Java process running under > the same UNIX user on the servers, then this and other weirdness can happen > because they share the same ticket cache file in /tmp. Someone does a > (re)login and another process concurrently accessing the cache gets garbage > or unexpected state. > > > On Mon, Mar 10, 2014 at 2:26 PM, Wei Tan <[email protected]> wrote: > > > Hi, > > > > We are running a HBase cluster in these settings and with kerberos > > enabled. > > HBase: 0.96.1.1 > > Zookeeper: 3.4.5 > > Hadoop: 1.1.1 > > > > > > We constantly put data into HBase and every several hours we get the > error > > below on a random region server; this error arises and the region server > > kills itself. > > > > ERROR: > > 2014-02-28 09:32:39,755 ERROR > [hconnection-0x116987ad-shared--pool1378-t9] > > security.UserGroupInformation: PriviledgedActionException > > as:XXXXXXXX@DOMAIN cause:javax.security.sasl.SaslException: GSS initiate > > failed [Caused by GSSException: No valid credentials provided (Mechanism > > level: The ticket isn't for us (35) - BAD TGS SERVER NAME)] > > > > > > > > We also tried with multiple version of kdc - all the way up to latest > > 1.12.1 - still see this error. What is weird is that most put gets > > processed successfully until this error occurs and kills the RS. > > > > Thanks, > > Wei > > --------------------------------- > > Wei Tan, PhD > > Research Staff Member > > IBM T. J. Watson Research Center > > http://researcher.ibm.com/person/us-wtan > > > > > -- > Best regards, > > - Andy > > Problems worthy of attack prove their worth by hitting back. - Piet Hein > (via Tom White) > -- Best Wishes! Yours, Zesheng
