> Nice analysis, thank you. Indeed as you guessed, option A is too expensive,
> it would bring us a few years back in terms of performance on large systems.
>
> In parallel, I suggest you have a look at recent fix ca686e3208
> ("BUG/MEDIUM: server: Properly check for streams before deletion"), as I
> suspect it might address the same problem. In this case we did an atomic
> op but only per thread-group, as is already done for a few other counters,
> which amortizes the cost.Thank you -- that confirms our own doubt about Proposal A, and ca686e3208 is exactly the fix. It covers the same two cases (failed-connection target and cookie persistence), and the per-thread-group nb_strm counter is the detail we missed: we reached for a global refcount and hit the cost you mention, whereas amortising it per tgroup keeps it affordable. So this supersedes both our proposals. Apologies for the mangled patches -- it was only a reference diff and ca686e3208 supersedes it anyway, so nothing to resend; we'll attach properly next time. Thanks again for the quick and detailed reply. Best regards, YoungKwang Lee
