add resource journal

doc/man1/flux-resource.rst

@@ -22,6 +22,7 @@ SYNOPSIS
 | **flux** **resource** **reload** [-f] [--xml] *path*
 
 | **flux** **resource** **acquire-mute**
+| **flux** **resource** **eventlog** [*-w* *EVENT*] [*-f*]
 
 DESCRIPTION
 ===========
@@ -356,6 +357,24 @@ Tell the resource module to stop sending RFC 28 ``resource.acquire`` responses
 to the scheduler.  This is used during Flux instance shutdown to avoid asking
 the scheduler to needlessly process OFFLINE updates.
 
+eventlog
+--------
+
+.. program:: flux resource eventlog
+
+Watch the resource journal, which is described in RFC 44.
+
+.. option:: -f, --follow
+
+  After printing the current eventlog, wait for new events and print
+  them as they arrive.
+
+.. option:: -w, --wait=EVENT
+
+  Wait for the specified event to be posted, print it, then quit.
+  The current set of valid events events is *restart*, *resource-define*,
+  *online*, *offline*, *drain*, *undrain*, *torpid*, and *lively*.
+
 OUTPUT FORMAT
 =============
 
@@ -520,3 +539,4 @@ FLUX RFC
 | :doc:`rfc:spec_22`
 | :doc:`rfc:spec_27`
 | :doc:`rfc:spec_29`
+| :doc:`rfc:spec_44`

src/cmd/flux-resource.py

@@ -720,6 +720,41 @@
     print(rset.encode())
 
 
+def print_events(events, follow, wait):
+    if not events and not follow and not wait:
+        return False
+    for entry in events:
+        print(json.dumps(entry))
+        if wait and entry["name"] == wait:
+            return False
+    return True
+
+
+def eventlog_continuation(f, follow, wait):
+    try:
+        payload = f.get()
+    except OSError as exc:
+        if exc.errno != errno.ENODATA:
+            raise
+        payload = None
+    if not payload or not print_events(payload["events"], follow, wait):
+        f.flux_handle.reactor_stop()
+    else:
+        f.reset()
+
+
+def eventlog(args):
+    """Show the resource eventlog"""
+    h = flux.Flux()
+    f = h.rpc(
+        "resource.journal",
+        nodeid=0,
+        flags=flux.constants.FLUX_RPC_STREAMING,
+    )
+    f.then(eventlog_continuation, args.follow, args.wait)
+    h.reactor_run()
+
+
 LOGGER = logging.getLogger("flux-resource")
 
 
@@ -1009,6 +1044,23 @@
     R_parser.add_argument("--from-stdin", action="store_true", help=argparse.SUPPRESS)
     R_parser.add_argument("--config-file", help=argparse.SUPPRESS)
 
+    eventlog_parser = subparsers.add_parser(
+        "eventlog", formatter_class=flux.util.help_formatter()
+    )
+    eventlog_parser.add_argument(
+        "-f",
+        "--follow",
+        action="store_true",
+        help="Display new events as they are posted",
+    )
+    eventlog_parser.add_argument(
+        "-w",
+        "--wait",
+        metavar="EVENT",
+        help="Display events until EVENT is posted",
+    )
+    eventlog_parser.set_defaults(func=eventlog)
+
     args = parser.parse_args()
     args.func(args)
 

src/modules/resource/monitor.c

@@ -112,6 +112,37 @@
     return idset_decode (members);
 }
 
+static int post_restart_event (struct monitor *monitor)
+{
+    struct idset *ranks = NULL;
+    char *ranks_str  = NULL;
+    char *online_str  = NULL;
+    const char *hostlist;
+    int rc = -1;
+
+    if (!(hostlist = flux_attr_get (monitor->ctx->h, "hostlist"))
+        || !(ranks = idset_create (monitor->size, 0))
+        || idset_range_set (ranks, 0, monitor->size - 1) < 0
+        || !(ranks_str  = idset_encode (ranks, IDSET_FLAG_RANGE))
+        || !(online_str  = idset_encode (monitor->up, IDSET_FLAG_RANGE))
+        || reslog_post_pack (monitor->ctx->reslog,
+                             NULL,
+                             0.,
+                             "restart",
+                             EVENT_NO_COMMIT,
+                             "{s:s s:s s:s}",
+                             "ranks", ranks_str,
+                             "online", online_str,
+                             "nodelist", hostlist) < 0)
+        goto done;
+    rc = 0;
+done:
+    ERRNO_SAFE_WRAP (free, online_str);
+    ERRNO_SAFE_WRAP (free, ranks_str);
+    idset_destroy (ranks);
+    return rc;
+}
+
 /* Post event 'name' with a context containing idset:s, where 's' is
  * the string encoding of 'ids'.  The event is not propagated to the KVS.
  */
@@ -361,6 +392,8 @@
                                  monitor) < 0)
             goto error;
     }
+    if (post_restart_event (monitor) < 0)
+        goto error;
 done:
     return monitor;
 error: