Skip to main content

Error: DDL statement interfered with this statement

  • March 18, 2014
  • 3 replies
  • 15 views

Naveen_Kumar
Forum|alt.badge.img+1
Hi,
when i was loading data continuously to vertica table using copy query, after some times i got this error-
" 08:16:03,557 ERROR VerticaProcessor:148 - Error in processing java.sql.SQLNonTransientException: [Vertica][VJDBC](3007) ERROR: DDL statement interfered with this statement

Caused by: com.vertica.support.exceptions.NonTransientException: [Vertica][VJDBC](3007) ERROR: DDL statement interfered with this statement"

Can anyone explain me root cause of this error?

3 replies

Bing_Wu
Forum|alt.badge.img+1
  • Participating Frequently
  • March 19, 2014
It is very likely there is another DDL statement running in parallel in another session. I have seen this type of error message myself when I was loading data in schema A while trying to drop a table in schema B. The catalog object in Vertica seems to be Global in scope. And DDL operations can easily lock each other out.

Nimmi_gupta
Forum|alt.badge.img
  • Participating Frequently
  • March 19, 2014
What version of vertica are you running? If it's V4.x, there was a bug and  resolved in 5.0 patch set.

Bing_Wu
Forum|alt.badge.img+1
  • Participating Frequently
  • March 20, 2014
In my case, I was running 6.1.x when I saw a message very much like in the OP (I don't remember whether it was from [VJDBC] or not). I was running a background (nohup) batch job that does bulk-load then bulk-insert:
copy A ... from local file;
copy B ... from local file;
insert into C 
select .. from A ;
insert into C
select ... from B;

There were dozen such statements in all, each takes at least minutes to complete. While running the script I realized that the disk space was below the 50% mark so I started another process to drop schemas: drop schema Y1 cascade;

Later I found out some tables in the loading job didn't get populated and went back to see the logs. That is when I saw the error message (from the loading script).

For various reasons I didn't create a support ticket (e.g. not a production instance, time-consuming to reproduce, etc.) Are you sure this issue was fixed in 5.0 and on ward?