Memory leak in Aurora Postgres since 3.9.1 about vertx-sql-client HOT 4 CLOSED

Thanks for looking into this!

Haven't tried DISCARD ALL; it might have a performance impact as it will flush cached query plans when connection is returned to the pool. Might try this if other options won't work.

If you use prepare, can you make sure the statement is closed properly?

The statement wasn't closed indeed. We were doing prepare(sql).coAwait().cursor(args) and then reading the cursor until it is exhausted, but never closed the result of prepare(sql). I changed this to:

flow<Row> {
    val preparedStatement = prepare(sql).coAwait()
    try {
        val cursor = preparedStatement.cursor(args)
        do cursor.read(pageSize).coAwait().forEach { emit(it) }
        while (cursor.hasMore())
    } finally {
        preparedStatement.close().coAwait()
    }
}

It looks promising so far, will see if we can keep connections open indefinitely. Based on the doc, I'm assuming that cursor doesn't have to be closed since we read it until the end.

from vertx-sql-client.

Fully removed max lifetime today - DB memory looks good. Thanks for your help!

from vertx-sql-client.

Have you tried using DISCARD ALL as suggested in the group discussion?

It would help to have some details about what is consuming memory, can you get that from your DBA?

In the 3.9.1 release notes, there is a link to #577
If you use prepare, can you make sure the statement is closed properly?

from vertx-sql-client.

Good news, keep us posted please.

from vertx-sql-client.