Labels

Android (1) bash (2) boost (2) C (34) C++ (2) cheatsheet (2) CLion (6) css (3) Debian (33) DL (17) Docker (2) Dreamweaver (2) Eclipse (3) fail2ban (4) git (5) GitHub (4) Hacking (3) html (8) http (1) iOS (1) iPad (1) IRC (1) Java (31) javascript (3) Linux (167) Mac (19) Machine Learning (1) mySQL (48) Netbeans (4) Networking (1) Nexus (1) OpenVMS (6) Oracle (1) Pandas (3) php (16) Postgresql (8) Python (9) raid (1) RedHat (14) Samba (2) Slackware (47) SQL (14) svn (1) tar (1) ThinkPad (1) Virtualbox (3) Visual Basic (1) Visual Studio (1) Windows (2)

Saturday, 7 March 2020

MySQL triggers

The below are examples of 2 triggers I set up on a database recently.

The first one was to evaluate a condition on inserting a new record, while the second one was to evaluate the same condition but this time on update.


Basically, I had several columns but this only concerned three of them, two of them were time stamps and one was a boolean (actually it was a tiny int, but we won't worry about this here)

We had fault_occuredfault_resolved and resolved.

The first trigger;


DELIMITER $$
CREATE TRIGGER before_insert
BEFORE INSERT
ON log FOR EACH ROW
BEGIN
IF NEW.fault_resolved>NEW.fault_occured THEN
SET NEW.resolved=1;
END IF;
END$$
DELIMITER ;

The second trigger;

DELIMITER $$
CREATE TRIGGER before_update
BEFORE UPDATE
ON log FOR EACH ROW
BEGIN
IF NEW.fault_resolved>OLD.fault_occured THEN
SET NEW.resolved=1;
END IF;
END$$
DELIMITER ;

I wasn't so sure about the OLD.fault_occured in the second one but it works when tested.


A few notes;

The table was log
We can see all triggers with;
SHOW TRIGGERS;
We can drop a trigger with;
DROP TRIGGER trigger_name;

I believe the above before_insert trigger can also be achieved with the following being run every time after an insert;

UPDATE log
SET resolved='1'WHERE (id=(SELECT MAX(id)) and fault_resolved > fault_occured);

Not so relevant, but here for completeness, the below is a construct of the table and al the sql to test the above triggers;


This tests that nothing happens to resolved when an insert is made with no fault_resolved is added;

INSERT INTO log (fault_occured, machine, module, fault, responsibility, username, details)VALUE ('2020-03-07 09:45','D','Sabre','Other','Engineering','hnisbet','');
This tests that resolved is properly set to 1 when an insert is made with a fault_resolved which is greater then the fault_occured;

INSERT INTO log (fault_occured, machine, module, fault, responsibility, username, details,fault_resolved)VALUE ('2020-03-07 09:10','D','Sabre','Other','Engineering','hnisbet','','2020-03-07 09:11');

The below is the sql to create the table that all of the above referred to;

CREATE TABLE log
(
    id             int AUTO_INCREMENT,
    fault_occured  timestamp  DEFAULT '0000-00-00 00:00:00' NOT NULL,
    machine        varchar(6)                               NOT NULL,
    module         varchar(20)                              NOT NULL,
    fault          varchar(50)                              NOT NULL,
    responsibility varchar(20)                              NOT NULL,
    resolved       tinyint(1) DEFAULT 0                     NOT NULL,
    username       varchar(20)                              NOT NULL,
    details        varchar(250)                             NULL,
    fault_resolved timestamp  DEFAULT '0000-00-00 00:00:00' NULL,
    CONSTRAINT log_id_uindex
        UNIQUE (id),
    CONSTRAINT log_module_fk
        FOREIGN KEY (module) REFERENCES modules (module)
            ON UPDATE CASCADE,
    CONSTRAINT log_users_username_fk
        FOREIGN KEY (username) REFERENCES users (username)
);

ALTER TABLE log
    ADD PRIMARY KEY (id);

NOTE:
The line fault_occured  timestamp  DEFAULT '0000-00-00 00:00:00' NOT NULL had to be manually changed as an after thought, as it originally was set to  NOT NULL and I guess as a result of this; DEFAULT was set to current timestamp, which was fine until I added the triggers, where I found on any update this would be changed to the current time, which was of course no good here.


No comments:

Post a Comment

Note: only a member of this blog may post a comment.