I'm trying to upgrade Liquibase from 3.3.1 to 3.10.3. Existing changesets are failing because varchar values being udpated are now inserting additional single quotes. Given the following statement in a changeset:
<update tableName="foo_table">
<column name="foo_id" type="VARCHAR(32)" value="'a'"/>
<where>bar_id REGEXP 'A'</where>
</update>
Liquibase seems to be adding additional single quotes in 3.10.3 so that values in the database are now stored with single quotes. Removing the single quotes from value in the changeset corrects the issue but will change the checksum on many of our changesets. I was wondering if there were any options available that will avoid us from needing to do that.
One option would be to create a new changelog (you can keep your old one, refer to both in order using includeAll in a master changelog). In the new changelog, you can just create your changes in sql.
This would allow you to write the change as sql exactly the way you want it, and liquibase would not have to render the sql based on xml.
Related
I have run into an issue with liquibase. So , I made a mistake in my sql commands in one of the changesets. Now whenever I try to upgrade my deployment , the upgrade is failing due to wrong checksum.
So I thought of doing the validCheckSum : ANY workaround and have run into this issue where it is not working. So my question is , is there any mistake in my syntax and How do I bypass the checksum check or what is the proper syntax for setting any in liquibase formatted sql.
This is what I have tried
--changeset darth: build-1.3
--validCheckSum: build-1.3:ANY
and it doesnt work.
Please help! Thanks!
The docs say:
Special value "1:any" will match to any checksum
Not sure why did you put "build-1.3:ANY"?
Also what you can do is:
Execute your code with latest liquibase changes onto an empty database
In DATABASECHANGELOG table look for the rows with changeset ids that you have been changing
Copy the MD5SUM value and paste it into your changeSet definition as a value of validCheckSum
It should run on any database with already executed changesets without errors.
It is worth mentioning that altering a change set is not recommended: by setting a validCheckSum value you will have "old" changeset already executed and "new" changes to this change set will be skiped. On an empty database your "new" changeset will be executed, so you will potentially have model inconsistencies.
I have a sequence in my database, which I generated through Liquibase. During refactoring, we decided that we didn't like the name that we gave it, and we would like to rename it, preserving all data that currently exists for it.
It seems possible to alter a sequence, but I'm not seeing anything about how to rename the sequence. Is there a way to do it, or a reasonable workaround?
(If it matters, I'm using Oracle SQL)
Although not documented, this refactoring is supported by Liquibase. Not sure what version this change was implemented in, but the class supporting the feature was commited on 30 Jan 2014. What's interesting though, is that the original issue is still unresolved.
Anyway, the refactoring is supposed to be working only on Oracle and Postgres. I've tested it on Oracle with Liquibase 3.4.1:
databaseChangeLog:
- changeSet:
id: change_set_id
author: me
dbms: oracle
changes:
- renameSequence:
oldSequenceName: old_name_seq
newSequenceName: new_name_seq
The above refactoring is in YAML format, but you could easily guess its XML counterpart.
On Oracle, this generates the following statement:
RENAME old_name_seq TO new_name_seq;
2 other supported parameters are catalogName and schemaName.
There is not currently a built-in refactoring to rename a sequence. If your database engine supports it, you could execute whatever methods are supported using a <sql> or <sqlFile> change.
You said you were using Oracle SQL. The RENAME statement allows for renaming a sequence. So your Liquibase script would look like this:
<sql>RENAME old_sequence_name TO new_sequence_name</sql>
I have question regarding liquibase.I am getting an error while inserting clob type data.String literal is too long.I have mentioned as follows
column name="help_item_text" type="clob" value="String too long more than 4000 charchaters"
But no luck still same.
Normally Liquibase uses standard SQL statements so that there is no difference between updateSql and update modes. For CLOB fields, that can run into problems when the total SQL length gets to be longer than the database's SQL parser can handle.
There is a valueClobFile attribute on column that allows you to save the long value to a file and then reference it from the changelog file. This gets the large value out of your changelog file and also tells Liquibase it needs to use a prepared statement.
<column name="help_item_text" type="clob" valueClobFile="String too long more than 4000 characters">
I am attempting to generate a change log against a MariaDB server. I am able to successfully generate a change log, however if I do a dropAll and update to validate that it is useful, there are multiple problems with it. I have tried using both the native mariadb and MySQL jdbc connectors and both experience these problems. I am also using liquibase 3.1.1.
The first is that there are deferrable and initiallyDeferred flags which are not supported by MySQL. The update command specifically calls these out as being invalid flags against MySQL.
Once I remove all of those references in the .xml, attempting to update runs into a sql syntax error because a double datatype is defined as DOUBLE(22) (in the xml). This is not a valid syntax for a double in mariadb or MySQL. They accept no parameters, or DOUBLE(m,d); my database is defined as default (no parameters).
Now its trying to create a table with an auto_increment but not specifying that the column is also a key in the create table statement; ie. it's missing the primaryKey constraint.
And I'm sure there are more problems in line as I work my way through the changelog (this is just changeset 116 out of 1500+).
Its almost as if liquibase is creating the changelog based on it thinking the db is a different type (postgres/oracle?).
Am I missing something?
You are right, the problem is that Liquibase is thinking it is an unknown database and doesn't know it is almost mysql. There is an issue open to add mariaDB support (https://liquibase.jira.com/browse/CORE-1411) but it hasn't been implemented quite yet.
The easiest work-around would be to add an extension:
Create a new liquibase.database.ext.MariaDBDatabase class in your codebase that extends liquibase.database.core.MySQLDatabase and override the isCorrectDatabaseImplementation(DatabaseConnection conn) method that returns true if conn.getDatabaseProductName() equals whatever the MariaDB jdbc driver is returning.
You may also want to override the getDefaultDatabaseProductName() to return "mariadb" instead of MySQL so you can differentiate it with contexts
This is what I have in the csv file:
CONTACT_TYP_CD,CONTACT_TYP_DESC,CREATE_DATE,CREATE_USER,UPDATE_DATE,UPDATE_USER
"ALL","Contact to be used for all communications","2014-03-14 00:00:00","CS_MAIN",null,null
This is how I load this file through liquibase:
<loadData file="src/main/resources/METAINF/install/seed_data/seed_contact_type.csv"
tableName="CONTACT_TYPE">
</loadData>
This is what liquibase uses to insert the data into oracle:
liquibase.exception.DatabaseException: Error executing SQL INSERT INTO CONTACT_TYPE (CONTACT_TYP_CD,CONTACT_TYP_DESC,CREATE_DATE,CREATE_USER,UPDATE_DATE,UPDATE_USER) VALUES ('"LL","Contact to be used for all communications","2014-03-14 00:00:00","CS_MAIN",null,null'): ORA-00947: not enough values
Can someone tell me what I am doing wrong? Thank you
Try removing the double quotes from your csv file, and define columns types, see the following test case as an example:
changeSet source
csv file
Another solution might be to enclose also the column titles in quotes, as seen here
you need to add the xml escape for double quote in the quotechar
quotchar="""
Your Problem here is the separator, not the quote. Liquibase seems to use semikolons by default, so you have to specify separator=",". Otherwise the whole line is considered a single value.