Error doing `SELECT` operations on Hive, `show databases` works fine about pyhive HOT 8 CLOSED

The error is on the Hive server side so it's hard to say what's going on. Are you able to run queries on beeline? In principle beeline also connects to HiveServer2 via Thrift. If beeline works, can you run set with both pyhive and beeline and diff the outputs?

from pyhive.

I did some more tests.

1. set works both on beeline and pyhive
2. doing a select on a table with name starting with a letter ('mytable') works fine both on pyhive and beeline
3. doing a select on a table with the name starting with a number ('2016_mytable') gives me the error, on different tables and different databases. The same query works on beeline

Can I provide anything else to help?

from pyhive.

Can you try backticking the names? It sounds like the two clients might be
using different Hive configuration. Hive has some config options that
affect identifier handling, e.g. hive.support.quoted.identifiers.

It would be useful to know if set returns the same output from both
clients, or if there's some config option that's different.

On Thu, Jan 14, 2016 at 10:51 AM, Ivan Marin [email protected]
wrote:

I did some more tests.

1. set works both on beeline and pyhive
2. doing a select on a table with name starting with a letter
   ('mytable') works fine both on pyhive and beeline
3. doing a select on a table with the name starting with a number
   ('2016_mytable') gives me the error, on different tables and different
   databases. The same query works on beeline

Can I provide anything else to help?

—
Reply to this email directly or view it on GitHub
#35 (comment).

from pyhive.

I can't backtick the names, but I tested with different tables on the same database and with very similar structure, with the only difference being the names with or without numbers on the beginning of the table names, and got the same results as reported before.

I got the two different sets and they're different. For example, the variable hive.support_quoted_identifies is getting u'hive.support.quoted.identifiers=column' from Pyhive and beeline. Which other variables are important?

Cheers

from pyhive.

Could you post the full diff of the set outputs? I'm not familiar with Tez so I'm not sure what would be relevant. At this point, I'm pretty sure this is an issue with configuration rather than an issue with the client.

from pyhive.

Unfortunately I can't post the diffs because of company policies. If you could guide me I can try to debug on my side to check what could be going wrong.

from pyhive.

I can't really help much, sorry. All I can say is to go through every single config and figure out what's different and whether it matters. This is probably a problem with the config that beeline and pyhive are passing up to the thrift server.

from pyhive.

Do you solved?
I also met

from pyhive.