Hi James, Thanks for your reply. My problem was resolved by setting phoenix.coprocessor.maxServerCacheTimeToLiveMs to 3 minutes and phoenix.upsert.batch.size to 10. I think, i can increase phoenix.upsert.batch.size to a higher value but haven't got opportunity to try that out yet.
Thanks, Anil Gupta On Thu, Jan 14, 2016 at 6:28 PM, James Taylor <jamestay...@apache.org> wrote: > Hi Anil, > This error occurs if you're performing an update that takes a long time on > a mutable table that has a secondary index. In this case, we make an RPC > before the update which sends index metadata to the region server which > it'll use for the duration of the update to generate the secondary index > rows based on the data rows. In this case, the cache entry is expiring > before the update (i.e. your MR job) completes. Try > increasing phoenix.coprocessor.maxServerCacheTimeToLiveMs in the region > server hbase-site.xml. See our Tuning page[1] for more info. > > FWIW, 500K rows would be much faster to insert via our standard UPSERT > statement. > > Thanks, > James > [1] https://phoenix.apache.org/tuning.html > > On Sun, Jan 10, 2016 at 10:18 PM, Anil Gupta <anilgupt...@gmail.com> > wrote: > >> Bump.. >> Can secondary index commiters/experts provide any insight into this? This >> is one of the feature that encouraged us to use phoenix. >> Imo, global secondary index should be handled as a inverted index table. >> So, i m unable to understand why its failing on region splits. >> >> Sent from my iPhone >> >> On Jan 6, 2016, at 11:14 PM, anil gupta <anilgupt...@gmail.com> wrote: >> >> Hi All, >> >> I am using Phoenix4.4, i have created a global secondary in one table. I >> am running MapReduce job with 20 reducers to load data into this >> table(maybe i m doing 50 writes/second/reducer). Dataset is around 500K >> rows only. My mapreduce job is failing due to this exception: >> Caused by: org.apache.phoenix.execute.CommitException: >> java.sql.SQLException: ERROR 2008 (INT10): Unable to find cached index >> metadata. ERROR 2008 (INT10): ERROR 2008 (INT10): Unable to find cached >> index metadata. key=-413539871950113484 >> region=BI.TABLE,\x80M*\xBFr\xFF\x05\x1DW\x9A`\x00\x19\x0C\xC0\x00X8,1452147216490.83086e8ff78b30f6e6c49e2deba71d6d. >> Index update failed >> at >> org.apache.phoenix.execute.MutationState.commit(MutationState.java:444) >> at >> org.apache.phoenix.jdbc.PhoenixConnection$3.call(PhoenixConnection.java:459) >> at >> org.apache.phoenix.jdbc.PhoenixConnection$3.call(PhoenixConnection.java:456) >> at org.apache.phoenix.call.CallRunner.run(CallRunner.java:53) >> at >> org.apache.phoenix.jdbc.PhoenixConnection.commit(PhoenixConnection.java:456) >> at >> org.apache.phoenix.mapreduce.PhoenixRecordWriter.write(PhoenixRecordWriter.java:84) >> ... 14 more >> >> It seems like i am hitting >> https://issues.apache.org/jira/browse/PHOENIX-1718, but i dont have >> heavy write or read load like wuchengzhi. I haven't dont any tweaking in >> Phoenix/HBase conf yet. >> >> What is the root cause of this error? What are the recommended changes in >> conf for this? >> -- >> Thanks & Regards, >> Anil Gupta >> >> > -- Thanks & Regards, Anil Gupta