Bug 2137412
Summary: | Make IDL clients connect to the new leader when a transaction fails due to leadership change. | ||
---|---|---|---|
Product: | Red Hat Enterprise Linux Fast Datapath | Reporter: | Mark Michelson <mmichels> |
Component: | openvswitch | Assignee: | Ilya Maximets <i.maximets> |
openvswitch sub component: | daemons and tools | QA Contact: | qding |
Status: | CLOSED WONTFIX | Docs Contact: | |
Severity: | unspecified | ||
Priority: | unspecified | CC: | ctrautma, fleitner, i.maximets, jhsiao |
Version: | FDP 22.E | ||
Target Milestone: | --- | ||
Target Release: | --- | ||
Hardware: | Unspecified | ||
OS: | Unspecified | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | If docs needed, set a value | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2024-04-23 21:32:02 UTC | Type: | Bug |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
Mark Michelson
2022-10-24 20:00:52 UTC
I'm not actually sure what this issue is about. I remember discussing it and I remember that it made some sense at the point of creation, but I think it is missing some context. For the problem statement in comment #0, the application supposed to re-check if transaction still applies after the transaction failed for the first time. It is necessary, because there is a chance that transaction will not fail and just be executed twice in case it doesn't violate database consistency. And the IDL clients do actually re-connect to a new leader if they are configured with leader-only connections. It would be nice if the transaction didn't fail, but that's the separate issue. It might be better to actually not re-connect to a new leader while there is an open transaction. In this case the client should receive a successful transaction reply once the transaction is committed. But that seems to be an opposite to what this BZ is asking for. I'll close this one for now. We may open an FDP Jira issue for keeping the connection to the old leader while there is an open transaction instead. |