why impala query execution ends before log shows 100% complete

[#1]

I am running the query in the above image.
Before the log shows "100% complete (128 out of 128) " the execution of the query stopped and showed that "Inserted 803974 rows".

 As you can see it stopped when the log is showing "74% Complete (95 out of 128)" ... and I am expecting more records to be inserted. 

where can I find the error/problem, if there is any?

[#2]

Hi @anis447 ,

Could you compare the actual row counts of the source and target tables? (like "select count(*) from v2_scopeFactTable;" and "select count(*) from dwh_db_atlas_jrtf.tblScopeFact;" ) to see if they match? The percentage values you describe seem to come from Hue, and they seem to be very different from the row count you mention, so I am not sure the two values are actually related.

Querying count(*) should give you the actual database row counts there.

Regards,

  - Laszlo

[#3]

Thanks @LaszloG 

But I don't know if you understood my question ... sorry if I was not clear.

First yes the query should insert more than 1M records in the created table, yet it inserted and stopped on around 800K.

Second what I am wondering and my question in another words is:
if you check the query log you will see that the "Status" of the query "FINISHED", yet the "Scan Progress" is "104 128 (81.25)% " as in the following image:

can you interpret that please?

[#4]

@anis447 Did you ever find an answer to this? We are hitting the same thing. Sometimes it finishes with < 100% and a query will return no data. Other times it will get to 100% and the same query will return data. 

Any information anyone has would be appreciated.

-Dave

[#5]

@PyMeH As this is an older post, you would have a better chance of receiving a resolution by starting a new thread. This will also be an opportunity to provide details specific to your environment that could aid others in assisting you with a more accurate answer to your question. You can link this thread as a reference in your new post. Thanks.

[#6]

Thanks Diana!

I just came on to post what we found for any future folks searching for a similar issue. It looks like this was potentially a client issue. We found this in the JDBC driver change log for 2.6.20 and it seemed to match the behavior we are seeing.

[00265194][IMPJ-606 ] When the server returns 0 rows, the driver stops fetching results.

We have updated our clients and have not been able to reproduce the issue. Presumably when the client incorrectly closes the connection, the server stops processing wherever it happens to be, which is why we see it stopping before the 100%. This is more of a symptom than the cause.