Re: [PATCH] [RFC] Decrease server health based on http responses / events

From: Willy Tarreau <>
Date: Thu, 10 Dec 2009 19:50:35 +0100

On Thu, Dec 10, 2009 at 07:17:30PM +0100, Krzysztof Ol?dzki wrote: (...)
> >"events" by itself might not be the proper name then. For instance, a
> >timeout is precisely a lack of event.
> It's a matter of definition, like if 0 is or is not a natural number.
> For me timeout *is* one of events, but no problem - we can choose a name
> that is intuitive for everyone.
> >Maybe simply "errors" ? The
> >other ones are not errors, just plain valid status codes after all.
> OK, but we are not only going to observing errors, but also succesfull
> connections to clear error counter. So maybe simple "layer4" (for tcp)
> and "layer7" (for http) might be OK?

If you see various uses of these levels, then yes, I agree.

> >yes, precisely. Another advantage would be that we could also
> >allow the statement on regular backend config (even defaults)
> >when it's supposed to be the same for all servers. It would
> >then be handled just like the "source" keyword : per-server,
> >then per-backend.
> Right. A backend/default value is important. However, I think we:
> - must use the same syntax for both backend's and server's configuration,
> - don't need to require to always specify all parameters
> - should keep current type of names: "type value" (not "to", "by",
> "after".
> So, I'll be more happy with "on-error fail-check" than with "by
> fail-check", etc.

OK, that's fine by me. I trust you to find an intuitive language :-)

Best regards,
Willy Received on 2009/12/10 19:50

This archive was generated by hypermail 2.2.0 : 2009/12/10 20:00 CET