nixos/zrepl: Improve tests (#416118)

nixos/zrepl: Improve tests to:

* Use a block device for zpool
* Have two virtual machines so that source->target replication works

nixos/tests/zrepl.nix

@@ -1,76 +1,178 @@
 {
   name = "zrepl";
 
-  nodes.host =
+  nodes = {
-    { pkgs, ... }:
+    source =
-    {
+      { nodes, pkgs, ... }:
-      config = {
+      {
-        # Prerequisites for ZFS and tests.
+        config = {
-        boot.supportedFilesystems = [ "zfs" ];
+          # Prerequisites for ZFS and tests.
-        environment.systemPackages = [ pkgs.zrepl ];
+          virtualisation.emptyDiskImages = [
-        networking.hostId = "deadbeef";
+            2048
-        services.zrepl = {
+          ];
-          enable = true;
-          settings = {
-            # Enable Prometheus output for status assertions.
-            global.monitoring = [
-              {
-                type = "prometheus";
-                listen = ":9811";
-              }
-            ];
-            # Create a periodic snapshot job for an ephemeral zpool.
-            jobs = [
-              {
-                name = "snap_test";
-                type = "snap";
 
-                filesystems."test" = true;
+          boot.supportedFilesystems = [ "zfs" ];
-                snapshotting = {
+          environment.systemPackages = [
-                  type = "periodic";
+            pkgs.parted
-                  prefix = "zrepl_";
+            pkgs.zrepl
-                  interval = "1s";
+          ];
-                };
+          networking.firewall.allowedTCPPorts = [ 8888 ];
+          networking.hostId = "deadbeef";
+          services.zrepl = {
+            enable = true;
+            settings = {
+              # Enable Prometheus output for status assertions.
+              global.monitoring = [
+                {
+                  type = "prometheus";
+                  listen = ":9811";
+                }
+              ];
+              # Create a periodic snapshot job for an ephemeral zpool.
+              jobs = [
+                {
+                  name = "snapshots";
+                  type = "snap";
 
-                pruning.keep = [
+                  filesystems."tank/data" = true;
-                  {
+                  snapshotting = {
-                    type = "last_n";
+                    type = "periodic";
-                    count = 8;
+                    prefix = "zrepl_";
-                  }
+                    interval = "10s";
-                ];
+                  };
-              }
+
-            ];
+                  pruning.keep = [
+                    {
+                      type = "last_n";
+                      count = 8;
+                    }
+                  ];
+                }
+                {
+                  name = "backup-target";
+                  type = "source";
+
+                  serve = {
+                    type = "tcp";
+                    listen = ":8888";
+
+                    clients = {
+                      "${nodes.target.networking.primaryIPAddress}" = "${nodes.target.networking.hostName}";
+                      "${nodes.target.networking.primaryIPv6Address}" = "${nodes.target.networking.hostName}";
+                    };
+                  };
+                  filesystems."tank/data" = true;
+                  # Snapshots are handled by the separate snap job
+                  snapshotting = {
+                    type = "manual";
+                  };
+                }
+              ];
+            };
           };
         };
       };
-    };
+
+    target =
+      { pkgs, ... }:
+      {
+        config = {
+          # Prerequisites for ZFS and tests.
+          virtualisation.emptyDiskImages = [
+            2048
+          ];
+
+          boot.supportedFilesystems = [ "zfs" ];
+          environment.systemPackages = [
+            pkgs.parted
+            pkgs.zrepl
+          ];
+          networking.hostId = "deadd0d0";
+          services.zrepl = {
+            enable = true;
+            settings = {
+              # Enable Prometheus output for status assertions.
+              global.monitoring = [
+                {
+                  type = "prometheus";
+                  listen = ":9811";
+                }
+              ];
+              jobs = [
+                {
+                  name = "source-pull";
+                  type = "pull";
+
+                  connect = {
+                    type = "tcp";
+                    address = "source:8888";
+                  };
+                  root_fs = "tank/zrepl/source";
+                  interval = "15s";
+                  recv = {
+                    placeholder = {
+                      encryption = "off";
+                    };
+                  };
+                  pruning = {
+                    keep_sender = [
+                      {
+                        type = "regex";
+                        regex = ".*";
+                      }
+                    ];
+                    keep_receiver = [
+                      {
+                        type = "grid";
+                        grid = "1x1h(keep=all) | 24x1h";
+                        regex = "^zrepl_";
+                      }
+                    ];
+                  };
+                }
+              ];
+            };
+          };
+        };
+      };
+  };
 
   testScript = ''
     start_all()
 
     with subtest("Wait for zrepl and network ready"):
-        host.systemctl("start network-online.target")
+        for machine in source, target:
-        host.wait_for_unit("network-online.target")
+          machine.systemctl("start network-online.target")
-        host.wait_for_unit("zrepl.service")
+          machine.wait_for_unit("network-online.target")
+          machine.wait_for_unit("zrepl.service")
 
-    with subtest("Create test zpool"):
+    with subtest("Create tank zpool"):
-        # ZFS requires 64MiB minimum pool size.
+        for machine in source, target:
-        host.succeed("fallocate -l 64MiB /root/zpool.img")
+          machine.succeed(
-        host.succeed("zpool create test /root/zpool.img")
+            "parted --script /dev/vdb mklabel gpt",
+            "zpool create tank /dev/vdb",
+          )
 
-    with subtest("Check for completed zrepl snapshot"):
+    # Create ZFS datasets
+    source.succeed("zfs create tank/data")
+    target.succeed("zfs create -p tank/zrepl/source")
+
+    with subtest("Check for completed zrepl snapshot on target"):
         # zrepl periodic snapshot job creates a snapshot with this prefix.
-        host.wait_until_succeeds("zfs list -t snapshot | grep -q zrepl_")
+        target.wait_until_succeeds("zfs list -t snapshot | grep -q tank/zrepl/source/tank/data@zrepl_")
+
+    with subtest("Check for completed zrepl bookmark on source"):
+        source.wait_until_succeeds("zfs list -t bookmark | grep -q tank/data#zrepl_")
 
     with subtest("Verify HTTP monitoring server is configured"):
-        out = host.succeed("curl -f localhost:9811/metrics")
+        out = source.succeed("curl -f localhost:9811/metrics")
 
         assert (
             "zrepl_start_time" in out
         ), "zrepl start time metric was not found in Prometheus output"
 
         assert (
-            "zrepl_zfs_snapshot_duration_count{filesystem=\"test\"}" in out
+            "zrepl_zfs_snapshot_duration_count{filesystem=\"tank/data\"}" in out
-        ), "zrepl snapshot counter for test was not found in Prometheus output"
+        ), "zrepl snapshot counter for tank/data was not found in Prometheus output"
   '';
 }