dleeyh Is this behavior only when using unix_timestamp() within transaction Or any other function in the transaction ?

were any other functions tried? is it just that one function, or do other temporal functions not work like from_unixtime(..) or now()? How about select substr('1', 1, 1) for example, to see if string functions have issues.

Instead of the unix_timestamp() function, I tried few other functions, such as min() and max(), and they did not cause the same issue.

those are aggregate functions, did you test any non-aggregate functions, did you test any other temporal functions?

It is very odd that unix_timestamp would have such a side effect.

Could you please test a string function and a temporal function if it has not been tested?

Note that many functions, including the temporal functions, are implemented in columnstore and don't use the ones in MariaDB code. MySQL has added the DATE_ITEM type to the optimizer since 5.1.73, and it breaks some of those implementations independently. Each needs to be tried to see what works and what doesn't. Fixes are usually straight forward.

okay, so something is strange here, because my test works, so this might not be related to unix_timestamp() or is side effect of previous test perhaps, or related to data types in lineitem perhaps?:
MariaDB [test]> set autocommit=0;
Query OK, 0 rows affected (0.00 sec)

MariaDB [test]> insert into t1 values(2);
Query OK, 1 row affected (0.02 sec)

MariaDB [test]> select unix_timestamp();
------------------

------------------

------------------
1 row in set (0.01 sec)

MariaDB [test]> commit;
Query OK, 0 rows affected (0.00 sec)

MariaDB [test]> select * from t1;
------

------

------
2 rows in set (0.01 sec)

in fact, why would select unix_timestamp(); even use columnstore? it uses no real tables...

So i created lineitem, and the test passes.

So this is some side effect of another portion of that test.

jswanhartIt is possible that the binaries that dleeyh tested this with on May 20th and the binaries that you are testing with are not same. Can you confirm the version of your binaries ? May be your binaries has updates that have fixed this behavior.

What we need to see is if the behavior reported by Daniel Lee is in the latest build - if not then we can close this, other wise analyze it further.

Build verified:

mscadmin> getsoft
getsoftwareinfo Mon Jun 6 22:13:55 2016

Name : mariadb-columnstore-platform Relocations: (not relocatable)
Version : 1.0 Vendor: MariaDB Corporation Ab
Release : 0 Build Date: Fri 03 Jun 2016 06:12:55 PM EDT
Install Date: Fri 03 Jun 2016 06:41:56 PM EDT Build Host: srvbuilder
Group : Applications Source RPM: mariadb-columnstore-1.0-0.src.rpm

This is no longed an issue in the build tested.

Tested in a 1UM-2PM stack.

MariaDB [mytest]> select unix_timestamp();
------------------

------------------

------------------
1 row in set (0.01 sec)

MariaDB [mytest]> insert into lineitem values(700000000,10000,1000,1,10,950.50,0.03,0.02,'N','O','1999-01-01','1999-01-01','1999-01-01','DELIVER IN PERSON','TRUCK','test test test test test');
Query OK, 1 row affected (2.15 sec)

MariaDB [mytest]> select unix_timestamp();
------------------

------------------

------------------
1 row in set (0.00 sec)

MariaDB [mytest]> commit;
Query OK, 0 rows affected (0.00 sec)

MariaDB [mytest]> desc lineitem;
-------------------------------------------------+

-------------------------------------------------+

-------------------------------------------------+
16 rows in set (0.00 sec)

MariaDB [mytest]> select * from lineitem where l_orderkey = 700000000;
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1 row in set (0.28 sec)

MariaDB [mytest]> quit
Bye

[root@s1um1 ~]# /usr/local/mariadb/columnstore/bin/viewtablelock
No tables are locked in the database.