Hi, First of all, your backend is labelled "oops" while in your error it says "Unused backend oopsy". (there is a Y at the end...), maybe a typo?
Second, you can use "restarts" and "backend.healthy" properties of a request to detect a failure and switch to your "oops" backend. cheers On Wed, Dec 1, 2010 at 2:57 PM, Cosimo Streppone <[email protected]> wrote: > Hi, > > This is varnish 2.0.4. I have a VCL where I define a backend like: > > backend nginx { > .host = "127.0.0.1"; > .port = "8080"; > } > > and then later in vcl_recv: > > set req.backend = nginx; > > Then I modified the VCL to add another backend, > to test it as downtime/maintenance service, so > that all requests coming in will just display a > "we're upgrading/we're offline" static page: > > backend oops { > .host = "1.2.3.4"; > .port = "80"; > } > > At the same time I modified req.backend to be "oops". > Now, I'm using only one backend. > If the backend that gets to be used in req.backend is > the last defined one, then everything's fine. > > If the used backend is any other than the last one, > I get this VCC error: > > Message from VCC-compiler: > Unused backend oopsy, defined: > (input Line 24 Pos 9) > backend oops { > --------#####-- > Running VCC-compiler failed, exit 1 > VCL compilation failed > > This came as a surprise... Is there a way around it? > > I'll probably setup a different "maintenance.vcl" > and alternate the VCL configs between "main.vcl" and "maintenance.vcl" > when I need to. > > Any better alternatives? > > -- > Cosimo > > _______________________________________________ > varnish-misc mailing list > [email protected] > http://www.varnish-cache.org/lists/mailman/listinfo/varnish-misc > _______________________________________________ varnish-misc mailing list [email protected] http://www.varnish-cache.org/lists/mailman/listinfo/varnish-misc
