I'm crafting Zapier task to insert entry in MySQL if there is no entry with specified unique key or do nothing.
I need to try to insert new row into some table, but if email of new entry is already INSERT will silently throw warning (due to IGNORE keyword).
INSERT IGNORE users(email, hashed_password)
VALUES ('<email>`, '<some_hashed_password>')
But in both scenarios my query is not returning anything and Zapier ends task with this message:
Question: Is there some way to have one complex SQL command that will combine INSERT and SELECT so with one query I will get some result set from from DB, not empty object or whatever INSERT returns?
P.S. This works in MySQL:
INSERT IGNORE reporting.users(`email`, `password`)
VALUES ("test#test.ts", "test");
SELECT * FROM reporting.users as u WHERE u.email = "test#test.ts";
but this consists of two queries and this doesn't work in Zapier.
This is an old question but I was grappling with the same issue today. In trying to find a solution I came across this qn and so when I found a solution / work-around I thought I'd do the decent thing and post back...
Based on the red "Bargle" error in their post I believe zmii must have been trying to use the MySQL custom search query. Zapier has to have an output from the query or it faults. I did some looking around and crafted my custom query thus:
SELECT IFNULL( (SELECT employee_id FROM timesheets.employees WHERE employee_id = <Step 6 | sheet_data_id> LIMIT 1) ,0) AS result;
based on the selected answer here.
So, my query will output an employee ID if it is found and 0 if not. I then inserted a Zapier PATHS step which I based on the output 'result' from my custom query. If the result is greater than 0 I update an existing record. If it is 0 then I insert a new record. I suspect I could use the custom query code to do the branching and updating/inserting but I didn't try that yet as I have other things to try first.
Edit
Actually I have had to revise this answer based on my conversation with Zapier here. I've retained the original answer but hidden it. The syntax works but only if the query is a SELECT query, it will not work for an INSERT or UPDATE query. See the discussion in the comments of the linked question for details. Essentially it is not possible to do an insert or update operation via the Zapier MySQL Custom Query step at this stage.
Related
I want to load data into Snowflake with Talend. I used tSnowflakeOutput with Upsert option because I want to insert data if not exists in Snowflake, or update rows if it exists. I used the primary key to identify the rows that already exist.
When I run my job, I have the following error:
Duplicate row detected during DML action
I am aware that the problem is due to a line that exists in Snowflake, I want to update the line but all I've got is this error.
do you have an idea why?
Please help :)
The Talend connector might be internally using the MERGE operation of Snowflake. As mentioned by #mike-walton, the error is reported because MERGE does not accept duplicates in the source data. Considering that its an insert or update if exists operation, if multiple source rows join to a target record, the system is not able to decide which source row to use for the operation.
From the docs
When a merge joins a row in the target table against multiple rows in the source, the following join conditions produce nondeterministic results (i.e. the system is unable to determine the source value to use to update or delete the target row)
A target row is selected to be updated with multiple values (e.g. WHEN MATCHED ... THEN UPDATE)
Solutions 1
One option as mentioned in the documentation can be to set the ERROR_ON_NONDETERMINISTIC_MERGE parameter. This will just pick an arbitrary source row to update from.
Solutions 2
Another option is to make it deterministic by using a MERGE query of the following form. This essentially does a de-duplication on the source table and lets you pick one of the duplicates as the preferred one for the update.
merge into taget_table t
using (
select *
from source_table
qualify
row_number() over (
partition by
the_join_key
order by
some_ordering_column asc
) = 1
) s
on s.the_join_key = t.the_join_key
when matched then update set
...
when not matched then insert
...
;
Doing this same thing in Talend may just require one to do a dedup operation upstream in the ETL mapping.
(NOTE: I know this is an error that's commonly asked about, but most of the time, the issue is in an INSERT statement. I couldn't find a question on this website where this error happened during an UPDATE.)
I have a table in MySQL (InnoDB / v. 5.7.19) called RESULTS which has, among others, two columns called TYPE and STATUS. Both are of type ENUM, with PASS, FAIL and IGNORE being the supported values in both. I'm trying to run this UPDATE statement on that table, using Workbench (also tried the same directly on the DB machine, using the mysql command):
update `RESULTS` set `TYPE`='IGNORE' where `STATUS`='IGNORE';
I'm getting this error:
Error Code: 1136. Column count doesn't match value count at row 1
Changing the single quotes to double quotes didn't help. I'm able to run this query successfully:
select count(`TYPE`) from `RESULTS` where `STATUS`='IGNORE';
I'm probably making a silly mistake here, but can anyone point out what's wrong with the UPDATE statement?
As requested I am posting it as an answer.
The error basically is self-explanatory like performing an operation on set of attributes but the values provided in the query are not enough. But in your case, you are performing an update operation with all attributes and their values and still, this error appears it may be a case that there is some trigger is registered for this table probably on before/after the event, If that is the case you need to update or remove that trigger if no needed.
I have read the other questions and all of the posted comments/answers to questions that are similar to this one. None of them seem to answer this question directly.
I am wanting to know if there is a way to either concatenate or place an INSERT, DELETE or UPDATE statement into a SELECT statement.
If so how? Could this be done by using a derived table function or a subSelect statement?
And example would be:
INSERT INTO Customer (name)
VALUES 'Test Name'
but place this into a SELECT statement
Example pseudo code:
SELECT *
FROM Customer
WHERE
Customer.name = (INSERT INTO Customer (name) VALUES 'Test Name')
The big thing here is getting the INSERT statement into the SELECT statement in an unconventional way.
Thank you and hopefully this will strike up some good conversation/ideas/results!
Reason for wanting to know this:
Our current DBMS (Fishbowl) does not allow us to use an UPDATE, DELETE or INSERT statement with is SQL compiler and we are wanting to mass alter our fields/clean up our tables.
To know if it is possible
You may have a select within an insert but not the other way
around.
A select within an insert statement could be used to copy values from a table to another. (You are just writing into a table after a read from another)
But an insert within a select doesn't make sense! What are you going to insert when you are running a select from a table(or just reading data)?(how is it possible to write when you are only allowed to read?)
Your situation, if you don't have access to run an insert then it doesn't matter where you put the statement, you just can't because you aren't allowed to!
But if you are talking about your query engine/database wrapper not allowing a direct insert, then its probably because it requires an application/program to insert data into to rather than just a query(as your engine doesn't already have the capability to perform that operation).
I have a query that runs everytime a user logins. Since this query regards information the user might have third-party updated recently I thought it would be a good idea to turn the user_id + information combo in the table unique. As so, everytime a user tried to save new information it would only save the one information I already didn't have. So, the first query being
INSERT INTO table VALUES ("1","cake"),("1","pie"),("1","bedsheets")
And as the user logins a second time and it being
INSERT INTO table VALUES ("1","cake"),("1","pie"),("1","bedsheets"),("1","chocolate")
It would only save ("1","chocolate") because (id,info) being an unique pair all other would not be inserted. I came upon the realization they all fail if only one fails. So my question is: is there any way to override this operation? Or do I have to query the db first to filter the information I already have? tyvm...
When you use the IGNORE Keyword, so every errors, in the execution are ignored. Example: if you have a duplicate or PRIMARY key error while executing a INSERT Statement, so it will ignored and the execution is not aborted
Use this:
I NSERT IGNORE INTO table VALUES ("1","cake"),("1","pie"),("1","bedsheets"),("1","chocolate");
I created a simple UPDATE AFTER INSERT trigger to sum two DECIMAL(10,2) fields and update a 3rd DECIMAL(10,2) field with the summed value.
The trigger code is:
delimiter |
CREATE TRIGGER calc_ttl_cost AFTER INSERT ON buybox_rec
FOR EACH ROW
BEGIN
UPDATE buybox_rec SET total_cost = (price+shipping_cost);
END;
|
I'm using phpMyAdmin to run the SQL commands, so I know that the trigger is being created successfully. Furthermore, if I run the UPDATE buybox_rec SET total_cost = (price+shipping_cost); SQL statement alone, it works as expected. (I've also tried the same code with back-ticks around the field names, but I wouldn't be writing all this if that worked)
The problem is that when I insert a new row, the trigger doesn't work and this error is thrown:
#1442 - Can't update table 'buybox_rec' in stored function/trigger
because it is already used by statement which invoked this stored
unction/trigger.
The error seems like some sort of recursive or circular reference problem, but I can't figure out where/what the problem is.
I also tried creating a stored procedure and calling it from within the trigger to try obtaining the same result, but I got the same error.
I checked a bunch of other SO questions related to MySQL UPDATE triggers and did some Googling, but here I am.
Is this out of the scope of MySQL? It seems like such a common and easy task to allow for.
Does anyone know how I accomplish this task autonomously? (AKA I don't want to hear about summing the fields after the fact via PHP, for example)
THanks for any help
The problem is that you're trying to modify the contents of a table which is already being used by the UPDATE + TRIGGER operation. This simply can't be done, but you have alternatives.
For example, if your meaningful data (or independent variables) are price and shipping cost while the total cost depends on them, you could keep only the first two in your table, and maybe have a very simple VIEW (something like SELECT price, shipping_cost, price+shipping_cost total_cost FROM buybox_rec, or whichever other fields you need) if you want to keep an eye at the total.