Also check the Hive Metastore log files. In general if the table has a large number of partitions incremental stats will have very large overhead in terms of metadata.
I would recommend running "compute stats bi_ full" then manually set the row count for newly added partitions whenever possible. On Wed, Jan 3, 2018 at 10:36 AM, Alexander Behm <[email protected]> wrote: > Thanks for the report. I have not seen this issue. Looks like the alter > RPC is rejected by the Hive Metastore. Maybe looking into the > Hive/Metastore logs would help. > > The SHOW CREATE TABLE output might also help us debug. > > On Wed, Jan 3, 2018 at 10:28 AM, Piyush Narang <[email protected]> > wrote: > >> Hi folks, >> >> >> >> I’m running into some issues when I try to compute incremental stats in >> Impala that I was hoping someone would be able to help with. I’m able to >> ‘compute stats’ in Impala on my smaller tables just fine. When I try >> computing stats incrementally for one of my larger tables, I seem to be >> running into this error: >> >> > compute incremental stats bi_ full partition >> (param1=0,day='2017-10-04',hour=00,host_platform='EU'); >> >> Query: compute incremental stats bi_full partition >> (param1=0,day='2017-10-04',hour=00,host_platform='EU') >> >> WARNINGS: ImpalaRuntimeException: Error making 'alter_partitions' RPC to >> Hive Metastore: >> >> CAUSED BY: InvalidOperationException: alter is not possible >> >> >> >> Looking at impalad.INFO and catalogd.INFO I don’t see any additional >> details. I verified that I’m the owner of the tables in HDFS. >> >> >> >> Has anyone run into this issue in the past? Any workarounds? >> >> >> >> Thanks, >> >> >> >> -- Piyush >> >> >> > >
