Member since
10-16-2013
307
Posts
77
Kudos Received
59
Solutions
My Accepted Solutions
Title | Views | Posted |
---|---|---|
11139 | 04-17-2018 04:59 PM | |
6112 | 04-11-2018 10:07 PM | |
3519 | 03-02-2018 09:13 AM | |
22105 | 03-01-2018 09:22 AM | |
2631 | 02-27-2018 08:06 AM |
02-02-2017
10:07 PM
@gaurang would you be open to sharing your CREATE TABLEs, CREATE VIEW and the query that has slow planning time? No need for the data, just that should be sufficient for us to understand better what's going on. Like Lars said, you are probably hitting IMPALA-4242 which explains the slow equivalence class computation, but I'd also like to understand the slow single-node planning time. Thanks!
... View more
01-19-2017
02:50 PM
1 Kudo
Are you running with HDFS HA (i.e. a setup with an active and a standby NN)? It is a known issue that Impala CTAS may not work with HDFS HA. As a workaround, CREATE+INSERT should work.
... View more
11-02-2016
05:41 PM
1 Kudo
Hi Maurin, both tables have to be scanned to observe SQL semantics. Otherwise, we would be changing the number of results coming out of your view. If you want the drop the second union operand, you could add a "WHERE a IS NOT NULL", and then the seocnd table will not be scanned. Alex
... View more
10-07-2016
09:28 PM
1 Kudo
Thanks for the report. The documentation is wrong. A REFRESH is always needed to pick up changes to table data that were made from tools other than Impala. I'll follow up with our docs team to get this fixed.
... View more
10-07-2016
06:12 PM
I agree completely. The workaround is not great. Dealing with Avro schemas in a more sensible fashion is definitely on our radar, but I cannot promise an ETA yet.
... View more
09-20-2016
08:03 AM
1 Kudo
Thanks for reporting this issue! Sorry for the trouble. I was able to reproduce the issue and can confirm this is a regression from previous Impala versions. I filed https://issues.cloudera.org/browse/IMPALA-4170 to track progress on the issue.
... View more
09-19-2016
09:52 AM
Thanks for following up and confirming that it works!
... View more
09-12-2016
05:20 PM
Thanks for following up! I'm pretty sure your table shoud work on more recent versions of Impala since we've fixed several Avro issues related to how schemas are defined. As a workaround, you could try the following things: 1. In your .avsc file make all fields nullable by specifying the types a a union of null and the type like this: type:["null", "int"] 2. Also specify corresponding matching column definitions in your CREATE TABLE, i.e. CREATE TABLE MI_FULL (col1 INT, col2 STRING, ) ROW FORMAT SERDE (the rest is exactly the same as before) Let me know if you have questions and whether those workarounds helped!
... View more
09-12-2016
01:38 PM
1 Kudo
Thank you for your colorful and honest feedback :). You've brought up a legitmiate concern around regressed behavior which is causing you pain. Very sorry. I agree that making the limit configurable is the best course of action in the short term. We'll take a look.
... View more