Update "write" function to work with pgx4 and higher

[hlcianfagna]

Summary of the changes / Why this is an improvement

Update to write function to work with pgx4

Closes: #88

Checklist

• Link to issue this PR refers to (if applicable):
• CLA is signed

[hammerhead]

I gave the patch a try but still ran into the originally reported error message.
In the pgx documentation I then found a note that using prepared statements is typically not necessary anymore, which seems to be new since v4:

However, this is rarely necessary because pgx includes an automatic statement cache by default

After replacing the prepared statement with a regular batched insert, I ran into error closing write batch: context deadline exceeded. A fix for that was to comment out this line:

cratedb-prometheus-adapter/crate.go

Lines 210 to 219 in 2f98328

	// The deadlock occurs when the batched queries to be sent are so large that the PostgreSQL
	// server cannot receive it all at once. PostgreSQL received some queued queries and starts
	// executing them. As PostgreSQL executes the queries it sends responses back. pgx will not
	// read any of these responses until it has finished sending. Therefore, if all network
	// buffers are full, pgx will not be able to finish sending the queries, and PostgreSQL will
	// not be able to finish sending the responses.
	//
	// -- https://github.com/jackc/pgx/blob/v3.6.2/batch.go#L58-L79
	//
	ctx, _ = context.WithTimeout(ctx, c.writeTimeout)

Only after that, I could see data arriving in the target table. The link in that context refers to pgx v3.6.2, and I cannot find that comment in the pgx source code anymore in v5. Sounds like that concern is obsolete as well?

[hlcianfagna]

I got this working now in my test environment, could you take another look at it @hammerhead ?

[hammerhead]

Looks good from a functional perspective. When running go run . -config.file ../config.yml on that branch, I can see data coming in successfully 👍.

[hlcianfagna]

I also looked at the possibility of adding a test case to test this function but it seems pgxmock would not currently support this ( pashagolub/pgxmock#152 )

[amotl]

Hi. Thank you so much for discovering the root cause of GH-88, and for submitting this patch and confirming it works well. I think it will be good to merge it without further ado.

Can I ask you to squash your commits, @hlcianfagna? Otherwise, I'll just do it, hoping that you don't have any objections about it?

[amotl]

Acknowledged, thank you. 💯