Implement Durable State handling with MySQL (#174) #363
Conversation
| UNIQUE KEY state_global_offset_uk (global_offset) | ||
| ); | ||
| CREATE INDEX state_tag_idx on durable_state (tag); | ||
| CREATE INDEX state_global_offset_idx on durable_state (global_offset); |
There was a problem hiding this comment.
[nit] add a newline at end of file
| // case MySQL => | ||
| // e shouldBe an[java.sql.SQLIntegrityConstraintViolationException] | ||
| case MySQL => | ||
| e shouldBe an[java.sql.SQLIntegrityConstraintViolationException] |
There was a problem hiding this comment.
Just for MySQL with an inconsistency with the line below (case Oracle)?
Or both for Oracle and SqlServer (not H2 and Postgres)
| case MySQL => slick.jdbc.MySQLProfile | ||
| case SqlServer => slick.jdbc.SQLServerProfile | ||
| case Oracle => slick.jdbc.OracleProfile | ||
| case _ => throw new UnsupportedOperationException(s"Unsupported <$s> for durableState.") |
There was a problem hiding this comment.
keep this - it causes no harm and is useful if we add a new db type and forget to support it here
There was a problem hiding this comment.
If a new DBMS is added, isn't it easier to have a non-exhaustive match reported by the IDE (and even some plugins) rather than a case _ that would hide it?
| } finally { | ||
| system.actorSelection("system/" + "pekko-persistence-jdbc-durable-state-sequence-actor").resolveOne().onComplete { | ||
| case Success(actorRef) => { | ||
| case Success(actorRef) => |
There was a problem hiding this comment.
I'm not a big fan of reformatting existing code that is not related to the PR
| @@ -0,0 +1,21 @@ | |||
| /* | |||
There was a problem hiding this comment.
new files should use the standard Apache header unless there are methods copied in from existing files - where we should use that file's header
eg https://github.com/apache/pekko-persistence-jdbc/blob/main/project/PekkoCoreDependency.scala#L1
|
|
|
I was fixing the tests when I realized that my code doesn't cover the case of "update the durable state record whose revision is -1 compared to the one I'm passing to the update" (WHERE clause in UPDATE This is something you can't do with the |
|
@marcopaggioro thanks for having a look at solving the issue. |
2 participants
#174
VARCHAR tagin the Postgres schema of thedurable_statetable ➡️VARCHARwithout a fixed length is not available in MySQL, becomeVARCHAR(255), as specified in the Oracle and SQLServer schema.state_payload BYTEAin the Postgres schema of thedurable_statetable ➡️BYTEAis not available in MySQL, but becomeLONGBLOB.global_offset BIGSERIALin the Postgres schema of thedurable_statetable ➡️BIGSERIAL(used asBIGINTwith auto-increment) is not available in MySQL, becomeBIGINT AUTO_INCREMENTand add aUNIQUE KEYtoglobal_offset.SequenceNextValUpdateronly for MySQL and useREPLACEquery