Hi Kirill, I look at the Oracle database, and it just throws an SQL timeout exception. The exception just canceled the query execution and doesn't roll back the transaction. Also, I checked the JDBC specification, and it indirectly says that the exception can be thrown, but it doesn't roll back a transaction (this is not an error; you can retry the operation). This behavior fits for our case, so.
Overall, I do not see any concern from the side of the community in the proposal. It's hard to offer any alternative based on the Ignite row-level locking protocol. Other issues arise during development, but the base is reflected here. I'll start creating tickets for implementation, there will be an umbrella one here[1]. [1]: https://issues.apache.org/jira/browse/IGNITE-28724 On Fri, May 22, 2026 at 3:46 PM ткаленко кирилл <[email protected]> wrote: > Hi, in the current implementation of the transactional protocol, the > proposal looks more or less. > > Tell me, what does the N in "WAIT <N> wait up to N" mean? > I would suggest limiting the request timeout. > -- Vladislav Pyatkov
