import ./make-test-python.nix ({ pkgs, ...} : let backend = { pkgs, ... }: { services.httpd = { enable = true; adminAddr = "foo@example.org"; virtualHosts.localhost.documentRoot = "${pkgs.valgrind.doc}/share/doc/valgrind/html"; }; networking.firewall.allowedTCPPorts = [ 80 ]; }; in { name = "proxy"; meta = with pkgs.stdenv.lib.maintainers; { maintainers = [ eelco ]; }; nodes = { proxy = { nodes, ... }: { services.httpd = { enable = true; adminAddr = "bar@example.org"; extraModules = [ "proxy_balancer" "lbmethod_byrequests" ]; extraConfig = '' ExtendedStatus on ''; virtualHosts.localhost = { extraConfig = '' Require all granted SetHandler server-status Require all granted BalancerMember http://${nodes.backend1.config.networking.hostName} retry=0 BalancerMember http://${nodes.backend2.config.networking.hostName} retry=0 ProxyStatus full ProxyPass /server-status ! ProxyPass / balancer://cluster/ ProxyPassReverse / balancer://cluster/ # For testing; don't want to wait forever for dead backend servers. ProxyTimeout 5 ''; }; }; networking.firewall.allowedTCPPorts = [ 80 ]; }; backend1 = backend; backend2 = backend; client = { ... }: { }; }; testScript = '' start_all() proxy.wait_for_unit("httpd") backend1.wait_for_unit("httpd") backend2.wait_for_unit("httpd") client.wait_for_unit("network.target") # With the back-ends up, the proxy should work. client.succeed("curl --fail http://proxy/") client.succeed("curl --fail http://proxy/server-status") # Block the first back-end. backend1.block() # The proxy should still work. client.succeed("curl --fail http://proxy/") client.succeed("curl --fail http://proxy/") # Block the second back-end. backend2.block() # Now the proxy should fail as well. client.fail("curl --fail http://proxy/") # But if the second back-end comes back, the proxy should start # working again. backend2.unblock() client.succeed("curl --fail http://proxy/") ''; })