Renato,
I did not change it personally , ever. My observation comes from the fact
that hive meta data stores location of a table (an hdfs path). I am using
mysql to store meta data. In my case, the mysql table 'SDS' has location
information, for example the table named "medex"  has location set in 'SDS'
as:

hdfs://hdfs.pww.rfiserve.net/user/hadoop/warehouse/reports/medex

If your namenode is running on a port other than the default. I doubt hive
will behave correctly unless meta data is made aware of it. I will wait for
some one to comment on a better way of solving this, playing with meta data
wont be safe and hence must be kept as last resort.

On Tue, Sep 28, 2010 at 4:54 PM, Renato Marroquín Mogrovejo <
renatoj.marroq...@gmail.com> wrote:

> Hi, what we did what Shrijeet said, we changed our hdfs port to 8020
> because we didn't know where to change it.
> It is a filthy workaround but as we are still testing Hive, we just let
> that one pass.
> Maybe someone knows a "dirty way" to change Hive's port, I would also be
> happy to try it.
> And Shrijeet what did you exactly chane in the meta store?
> Thanks!
>
>
> Renato M.
>
>
> 2010/9/28 Shrijeet Paliwal <shrij...@rocketfuel.com>
>
> Here is what might be happening:
>>
>> 1> You were running namenode on port 8020
>> 2> Your hive meta data was set based on that. (Table location on hdfs)
>> 3> You changed the namenode port (but hive meta data doesnt know about it)
>> 4> 'show tables' is still trying to find tables in
>> hdfs:<host>:8020/<table_path>
>>
>> I dont know a clean way to change the hive meta data if namenode host or
>> port changes. A kludge is to modify hive meta data as stored in mysql (or
>> where ever else hive meta data is).
>>
>> On Tue, Sep 28, 2010 at 3:58 PM, Matt Tanquary 
>> <matt.tanqu...@gmail.com>wrote:
>>
>>> This worked at one time, but I now I'm having an issue:
>>>
>>> I have a basic python script for testing python/hive. The script just
>>> does a few simple things:
>>>
>>> -show tables
>>> -describe [a table]
>>> -select * from [a table] limit 15
>>>
>>> The show tables and describe sections are functioning. However, when
>>> it gets to the select command, then the hive server begins to issue:
>>>
>>> 10/09/28 13:38:24 INFO ipc.Client: Retrying connect to server:
>>> mysvr/15.6.84.51:8020. Already tried 0 time(s).
>>> 10/09/28 13:38:25 INFO ipc.Client: Retrying connect to server:
>>> mysvr/15.6.84.51:8020. Already tried 1 time(s).
>>> 10/09/28 13:38:26 INFO ipc.Client: Retrying connect to server:
>>> mysvr/15.6.84.51:8020. Already tried 2 time(s).
>>> 10/09/28 13:38:27 INFO ipc.Client: Retrying connect to server:
>>> mysvr/15.6.84.51:8020. Already tried 3 time(s).
>>> 10/09/28 13:38:28 INFO ipc.Client: Retrying connect to server:
>>> mysvr/15.6.84.51:8020. Already tried 4 time(s).
>>> 10/09/28 13:38:29 INFO ipc.Client: Retrying connect to server:
>>> mysvr/15.6.84.51:8020. Already tried 5 time(s).
>>> 10/09/28 13:38:30 INFO ipc.Client: Retrying connect to server:
>>> .mysvr/15.6.84.51:8020. Already tried 6 time(s).
>>> 10/09/28 13:38:31 INFO ipc.Client: Retrying connect to server:
>>> mysvr/15.6.84.51:8020. Already tried 7 time(s).
>>> 10/09/28 13:38:32 INFO ipc.Client: Retrying connect to server:
>>> mysvr/15.6.84.51:8020. Already tried 8 time(s).
>>> 10/09/28 13:38:33 INFO ipc.Client: Retrying connect to server:
>>> mysvr/15.6.84.51:8020. Already tried 9 time(s).
>>>
>>> The problem I believe is the port that it's trying to connect to. My
>>> hdfs namenode is serviced on port 54310.
>>>
>>> I tried adding the following setting to hive-site.xml:
>>> <name>fs.default.name</name> <value>hdfs://mysvr:54310</value>
>>>
>>> Any other suggestions?
>>>
>>> Thanks!
>>> -M@
>>>
>>
>>
>

Reply via email to