> I'd prefer to prioritize user intuition, since putting the storage port in > the host field does not reflect how end-user traffic works in production. I agree w/Jaydeep's perspective here.
On Mon, Aug 18, 2025, at 2:21 PM, Francisco Guerrero wrote: > Yeah, I think this is a good suggestion. We can definitely think of > enhancing the audit log entry. > > On 2025/08/18 13:27:03 Paulo Motta wrote: > > off-topic, but it would be great to start using actual host IDs as > > identifiers everywhere rather than perpetuating IP:ports which are not > > actual node ids :( but I believe this would be an utopia :D > > > > On Mon, Aug 18, 2025 at 8:34 AM Mick <m...@apache.org> wrote: > > > > > > > > > > > > > > > > > > I'd like to bring up for discussion the host field in audit logs, which > > > currently shows > > > > the storage port (e.g., 192.168.1.100:7000) instead of the native port > > > users expect to see. > > > > > > > > Background: > > > > - Original implementation[1] used storage port for consistency with > > > other subsystems > > > > - CASSANDRA-7544[2] enabled multiple instances per IP, making storage > > > port the > > > > standard differentiator > > > > - This creates confusion for users reviewing client audit logs who > > > expect to see the > > > > native port (i.e 9042) > > > > > > > > Arguments: > > > > - Keep storage port: Consistent with gossip/repair/logs, maintains > > > existing behavior > > > > - Switch to native port: More intuitive for audit log analysis, > > > matches user expectations > > > > > > > > Considerations: > > > > 1. Should audit logs prioritize consistency with internal systems or > > > user intuition? > > > > 2. Would this change break existing tooling? > > > > 3. Should the change only land in trunk, or backport to all branches > > > up to 4.0? > > > > > > > > > Out of curiosity… > > > Is this host field used for anything other than identification ? > > > If it's purely an identifier field without need to the format, could it be > > > in the form "192.168.1.100:9042[7000]" ? > > >
