I am dealing with a table in hive which does not have partitions and with input format as textinputformat. This is not an external table and I create it using "Create table as select" template.
I use the alter table statement to rename the table as given below:
ALTER TABLE testdb.temptable RENAME TO testdb.newtable;
I get the following error:
Error: Error while compiling statement: FAILED: ParseException line 1:32 mismatched input 'RENAME' expecting KW_EXCHANGE near 'temptable' in alter exchange partition (state=42000,code=40000)
Closing: org.apache.hive.jdbc.HiveConnection
I see it is a bug in hive. I use the version:
Hive 0.12.0-cdh5.1.4
How do i go about fixing this issue. Thanks in advance for the help!
It's not exactly a bug, just a side effect of Open Source when it's done by a motley crew of people all around the world with no "product owner" and no incentives to use a common programming style (or run extensive regression tests, or <insert your complaint here>).
Aaaaaaah, now that it's said, I feel better... Let's get to the point.
In HiveQL the alter command does not use the same semantics as create or select; specifically, you cannot use the "ALTER DATABASE.TABLE" notation. If you try, then the HQL parser just fails with a queer error message, as you can see by yourself.
That's the way it is. You must type a use command first, then your alter command with just the table name. Yes, it sucks. But that's the way it is. And I see no reason why it should improve any time soon.
[Update Jun-2017] looks like ALTER finally supports the DB.TABLE syntax, on recent Cloudera distro (tested on CDH 5.10 with Hive 1.1.0 - but since they usually include a number of back-ports in their distro, maybe it's a feature of Hive 1.2+)
I have similar error message it is gone after using alternative syntax, selecting schema and reference table by the short name:
USE mydb;
ALTER TABLE mytable RECOVER PARTITIONS;
Related
I want to run sql like:
CREATE TEMPORARY TABLE tmptable AS SELECT * FROM redshift_table WHERE date > #{date};
I can run this sql in command line in Redshift, but if I run it in my program, it doesn't work correctly. When I change CREATE TEMPORARY TABLE to CREATE TABLE it works correctly.
I am using mybatis as OR mapper and driver is:
org.postgresql.Driver
org.postgresql:postgresql:9.3-1102-jdbc41
What's wrong?
I am assuming the #date is an actual date in your actual query.
Having said that, there is not reason this command doesnt work, its as per the syntax listed here,
http://docs.aws.amazon.com/redshift/latest/dg/r_CREATE_TABLE_AS.html
Have you tried posting it on AWS Redshift forums, generally they are quite responsive. Please update this thread too if you find something, this is quite an interesting issue, thanks!
I'm using a Win XP box with BDE Administrator and Access 2007 installed. I'm able to open and perform select queries on existing Paradox tables without problem but have some very strange behavior when attempting INSERT/UPDATE. I can even create a new Paradox table and it has the same behavior. Here is sample code:
' create new table
conObj.Execute "CREATE TABLE test (id INT, comment VARCHAR(30))"
' first insert works fine
conObj.Execute "INSERT INTO test VALUES (1, 'something')"
' second insert fails for unknown reason
conObj.Execute "INSERT INTO test VALUES (2, 'something else')"
I've tried using Jet 4.0, MS Access Paradox driver, and native Paradox driver connection strings but all yield the same result. On the second insert statement it throws an error:
Operation must be an updateable query
I've read numerous posts in forums and pages on help sites that tell me this error is caused by a file permissions issue. The account running this script is part of the Administrator group and I've changed file permissions to allow the Everyone group Full Control of the db file but this changes nothing.
This page put out by Micrsoft Support did not fix the problem: http://support.microsoft.com/kb/175168
Additionally, I can create a new table but any time I try to create a PRIMARY KEY or UNIQUE field I get an error message that says:
"Index_[random characters] is not a valid name."
try
"CREATE TABLE test (id INT, comment VARCHAR(30), primary key(id))"
I don't know much about Paradox databases but this has indeed been a learning experience. Even though I have a table file called table.db that's not enough to store more than a single row of data. I also need several other files to insert or update a paradox database:
table.DB
table.PX
table.VAL
table.XG0
table.XG1
table.YG0
table.YG1
I was nosing around in another program that generates paradox databases and found when I copied a blank database from it along with these other files it generated I was able to insert and update without problems. I have no idea what these files are for or why they need to be present to insert or update but having them present fixed my issue.
I apologize for posting a question that seems to have been asked numerous times on the internet, but I can't quite fix it for some reason.
I was trying to populate some tables using Oracle's magical sqldr utility, but it throws an ORA-01775 error for some reason.
Everywhere I go on Google, people say something along the lines of: "Amateur, get your synonyms sorted out" (that was paraphrased) and that's nice and all, but I did not make any synonyms.
Here, the following does not work on my system:
SQLPLUS user/password
SQL>CREATE TABLE test (name varchar(10), id number);
SQL>exit
Then, I have a .ctl file with the following contents:
load data
characterset utf16
infile *
append
into table test
(name,
id
)
begindata
"GURRR" 4567
Then I run this command:
sqlldr user#localhost/password control=/tmp/controlfiles/test.ctl
The result:
SQL*Loader-702: Internal error - ulndotvcol: OCIStmtExecute()
ORA-01775: looping chain of synonyms
Part of test.log:
Table TEST, loaded from every logical record.
Insert option in effect for this table: APPEND
Column Name Position Len Term Encl Datatype
------------------------------ ---------- ----- ---- ---- ---------------------
NAME FIRST 2 CHARACTER
ID NEXT 2 CHARACTER
SQL*Loader-702: Internal error - ulndotvcol: OCIStmtExecute()
ORA-01775: looping chain of synonyms
And, if I try to do a manual insert:
SQL> insert into test values ('aa', 56);
1 row created.
There is no problem.
So, yeah, I am stuck!
If it helps, I am using Oracle 11g XE on CentOS.
Thanks for the help guys, I appreciate it.
EDIT:
I kind of, sort of figured out part of the problem. The problem was that somewhere along the line, maybe during a failed load or something, Oracle had given itself corrupt views and synonyms.
The affected views were: GV_$LOADISTAT, GV_$LOADPSTAT, V_$LOADISTAT and V_$LOADPSTAT. I am not quite sure why the views got corrupt, but recompiling them resulted in compiled with errorserrors. The synonyms used in the queries themselves were corrupt, namely the gv$loadistat, gv$loadpstat, v$loadistat and v$loadpstat synonyms.
I wasn't sure about why this was happening and I didn't quite understand anything. So, I decided to drop the synonyms and recreate them. Unfortunately, I couldn't recreate them, as the view they pointed to (there is a bit of weird recursion going on here...) was corrupt. These views were the aforementioned GV_$LOADISTAT and other views. In other words, the synonyms pointed to the views that used those synonyms. Talk about a looping chain.
So...I recreated the public synonyms but instead of specifying the view as GV_$LOADISTAT, I specified them as sys.GV_$LOADISTAT. e.g.
DROP PUBLIC synonym GV$LOADISTAT;
CREATE PUBLIC synonym GV$LOADISTAT for sys.GV_$LOADISTAT;
Then, I recreated the user views to point to those public synonyms.
CREATE OR REPLACE FORCE VIEW "USER"."GV_$LOADISTAT" ("INST_ID", "OWNER", "TABNAME", "INDEXNAME", "SUBNAME", "MESSAGE_NUM", "MESSAGE")
AS
SELECT "INST_ID",
"OWNER",
"TABNAME",
"INDEXNAME",
"SUBNAME",
"MESSAGE_NUM",
"MESSAGE"
FROM gv$loadistat;
That seemed to fix the views/synonyms. Yeah, it is a bit of a hack, but it somehow worked. Unfortunately, this was not enough to run SQL Loader. I got a table or view does not exist error.
I tried granting more permissions to my regular user, but it didn't work. So, I gave up and ran SQL Loader as sysdba. It worked! It is not a good thing to do, but it is a development only system made for testing purposes, so, I didn't care.
I could not repeat your looping synonym chain error, but it appears the control file needed a bit of work, at least for my environment.
I was able to get your example to work by modifying it thusly:
load data
infile *
append
into table test
fields terminated by "," optionally enclosed by '"'
(name,
id
)
begindata
"GURRR",4567
I'm trying to convert some Informix ESQL to Oracle Pro*C. In the existing Informix code the "SERIAL" data type was used to indicate automatically incrementing columns. According to the Oracle documentation, the Oracle Migration Workbench for Informix should be able to handle this, and it explains that it converts the "SERIAL" data type into a "NUMBER" with an associated Oracle sequence and trigger. However, when trying to run the tool it simply replaces the word "SERIAL" with "ERROR(SERIAL)", so I've been trying to manually add in the trigger/sequence.
Their example here: http://docs.oracle.com/html/B16022_01/ch2.htm#sthref112 shows a way that this can be done. The sequence appears to be fairly straight forward, however when trying to create a trigger like so:
CREATE TRIGGER clerk.TR_SEQ_11_1
BEFORE INSERT ON clerk.JOBS FOR EACH ROW
BEGIN
SELECT clerk.SEQ_11_1.nextval INTO :new.JOB_ID FROM dual; END;
The Pro*C preprocessor picks up the "CREATE" keyword here, and decides that I'm not allowed to use the host variable ":new.JOB_ID", because host variables cannot be used in conjunction with "CREATE" statements.
My question is, is there some way to create a trigger that links an Oracle sequence to a particular column without using a host variable to specify the column name? The Oracle documentation seems to indicate that their migration tool should be able to cope, which means there must be some way of doing this. However all the examples of the trigger use that I have found all use the host variable which causes the preprocessor to complain.
Thank you for your time.
(Note: I've used the trigger/sequence/column names from the example in the Oracle documentation in the example above.)
I managed to resolve the issue by using an "EXEC SQL EXECUTE IMMEDIATE" statement.
char sql_buf[4096+1];
snprintf(sql_buf, 4096, <sql>);
EXEC SQL IMMEDIATE :sql_buf;
This bypasses the preprocessor and therefore allows the statement through without complaint.
It is impossible to create a trigger that links an Oracle sequence to a particular column without using a "host variable" to specify the column name. By the way it isn't "host variable" - just reference. The same trigger may fire on update and insert for example, so you have to specify what you are referencing: new or old variables. You can do it in MS-SQL but not in Oracle.
Can any one give me the syntax to truncate a table in IBM DB2.
I m running the following command: truncate table tableName immediate;
The eror is DB2
SQLCODE=-104, SQLSTATE=42601, SQLERRMC=table;truncate ;JOIN , DRIVER=3.50.152
Message: An unexpected token "table" was found following "truncate ". Expected tokens may include: "JOIN ".. SQLCODE=-104, SQLSTATE=42601, DRIVER=3.50.152
The syntax matches the one specified in the reference docs of IBM : http://publib.boulder.ibm.com/infocenter/dzichelp/v2r2/index.jsp?topic=/com.ibm.db29.doc.sqlref/db2z_sql_truncate.htm
There is a great article about truncating, here is the Gist of DB2 stuff
Almost follows the standard.(since version 9.7)
DB2 requires that the IMMEDIATE keyword be added the the ordinary TRUNCATE TABLE statement, e.g.:
TRUNCATE TABLE someschema.sometable IMMEDIATE
TRUNCATE TABLE must be the first statement in a transaction. A transaction starting with TRUNCATE TABLE may include other statements, but if the transaction is rolled back, the TRUNCATE TABLE operation is not undone.
DB2s TRUNCATE TABLE operation has a number of optional arguments, see the documentation for more on this; especially, the REUSE STORAGE argument may be important for ad-hoc DBA tasks.
In DB2 versions < 9.7, you may abuse the IMPORT statement. Unfortunately, you need to know which operating system the command is executed from for this to work:
On unix-like systems:
IMPORT FROM /dev/null OF DEL REPLACE INTO tablename
On Windows:
IMPORT FROM NUL OF DEL REPLACE INTO tablename
IMPORT cannot be abused in all contexts. E.g., when working with dynamic SQL (from Java/.NET/PHP/...—not using the db2 command line processor), you need to wrap the IMPORT command in a call to ADMIN_CMD, e.g.:
CALL ADMIN_CMD('IMPORT FROM /dev/null OF DEL REPLACE INTO tablename')
IMPORT seems to be allowed in a transaction involving other operations, however it implies an immediate COMMIT operation.
The ALTER TABLE command may also be abused to quickly empty a table, but it requires more privileges, and may cause trouble with rollforward recovery.
This was taken from the website:
http://troels.arvin.dk/db/rdbms/#bulk-truncate_table-db2
If you are using DB2 for AS400, IMMEDIATE TRUNCATE TABLE will NOT work. The equivallent work around is to either:
DELETE FROM [tableName] then if it is an auto increment equivalant column, run:
ALTER TABLE ALTER COLUMN RESTART WITH 1
OR the faster (most efficient way)
Pass a command to the system to clear out the Physical File
Java syntax:
CommandCall command = new CommandCall(new AS400(AS400SystemName, AS400JavaUser, AS400JavaPwd));
try {
command.run("CLRPFM FILE(as400SchemaName/" + tableName + ")");
Which version of DB2 are you using? The truncate table command was introduced in DB2 v9 (at least on the mainframe, which appears to be what you're asking about based on the link).
You may have to resort to the delete from option although this article gives a stored procedure way of doing it in DB2 v8.
use truncate 'table_name' immediate
This is the exact reference documentation available for TRUNCATE in DB2 from 9.7 version
DB2 Reference for TRUNCATE
DB2 on z/OS V10
Empty one of your tables: truncate table; followed by commit work. Ex. truncate temp;
Someone else table: truncate owner.table Ex: truncate student.work ;
I have not tried this on a linked DB2. I do not know if truncate node2.student.work; is good.
SQL for creating list of tables automatically. Substring (substr) used because column width for table name and creator are sooo long. Your values may be different.
select 'truncate table '||substr(creator,1,9)||'.'||substr(name,1,20)
from sysibm.systables
where creator = 'Student';
in Java make sure it's the first statement in the transaction