Fixing small issues

Closes #68
Closes #69 (except the 'edit' part which I cannot find)

Also updates the README.md wrt how it's deployed.

README.md

@@ -3,15 +3,33 @@
 The user setup behind [stats.c4dt.org](https://stats.c4dt.org).
 
 At root, it contains a bunch of script to run measurement, each regularly
-started by an user systemd timer, found in `.config`.
+started by a user systemd timer, found in `.config`.
 The timers output the results in a graphite database and shown in a grafana.
 Both are handle by the root `docker-compose.yaml`.
 To add or modify grafana's panels, look in `data/grafana`.
 
-Deployment is done via Github Actions, which do:
-
-- rsync of the repo to the home directory of the user on the stats' server
-- stop all timers of the user
-- start timers in the repo
+Deployment is done via ansible: [stats role](https://github.com/c4dt/ansible-config/tree/main/playbooks/roles/stats)
 
 If you need help with systemd, there is a [Systemd Cheatsheet](README.systemd.md)
+
+## Testing
+
+To test the new values, the easiest way is to do the following:
+
+- ssh to the stats-server, then
+
+```bash
+sudo -iu stats
+git pull
+git checkout your_branch
+make
+```
+
+Once you've finished testing, don't forget to
+
+```bash
+git checkout main
+make
+```
+
+And then re-apply ansible.

dashboards-to-provision/servers.dashboard.py

@@ -22,6 +22,8 @@
 from grafanalib.influxdb import InfluxDBTarget
 
 
+# the 'every' in the 'aggregateWindow' will always aggregate to 800 points,
+# no matter the range being shown.
 def target(server_name: str) -> InfluxDBTarget:
     """Generate Target for disk usage on given server"""
     return InfluxDBTarget(
@@ -33,7 +35,8 @@ def target(server_name: str) -> InfluxDBTarget:
             |> filter(fn: (r) => r.host == "{server_name}")
             |> filter(fn: (r) => r._field == "used_percent")
             |> keep(columns: ["_time", "_value", "path"])
-            |> last()
+            |> aggregateWindow(every: duration(v:(uint(v: v.timeRangeStop) - 
+               uint(v: v.timeRangeStart))/uint(v: 800)), fn: mean)
         """,
     )
 
@@ -66,46 +69,46 @@ def target(server_name: str) -> InfluxDBTarget:
         ]
     ),
     panels=[
-        GaugePanel(
-            title="$server: disk usage",
-            gridPos=GridPos(h=8, w=32, x=0, y=0),
-            targets=[target("$server")],
-            dataSource=DATA_SOURCE,
-            calc="lastNotNull",
-            thresholds=[
-                Threshold(index=0, color="green", value=0.0),
-                Threshold(index=1, color="orange", value=75.0),
-                Threshold(index=2, color="red", value=90.0),
-            ],
-            format=UNITS.PERCENT_FORMAT,
-            thresholdMarkers=False,
-            repeat=Repeat(direction="v", variable="server"),
-        )
-    ]
-    + [
-        # one graph/serie per host to actually generate alerts
-        Graph(
-            title=f"{host}: trigger disk usage alert",
-            gridPos=GridPos(h=0, w=0, x=0, y=0),  # invisible
-            targets=[target(host)],
-            dataSource=DATA_SOURCE,
-            nullPointMode=NULL_AS_NULL,
-            alert=Alert(
-                name=f"{host} disk nearly full",
-                message=f"{host} disk nearly full",
-                frequency="10s",
-                gracePeriod="1m",
-                alertConditions=[
-                    AlertCondition(
-                        target=target(host),
-                        timeRange=TimeRange("1m", "now"),
-                        evaluator=GreaterThan(90),
-                        operator=OP_AND,
-                        reducerType=RTYPE_LAST,
-                    ),
-                ],
-            ),
-        )
-        for host in HOSTS
-    ],
+               GaugePanel(
+                   title="$server: disk usage",
+                   gridPos=GridPos(h=8, w=32, x=0, y=0),
+                   targets=[target("$server")],
+                   dataSource=DATA_SOURCE,
+                   calc="lastNotNull",
+                   thresholds=[
+                       Threshold(index=0, color="green", value=0.0),
+                       Threshold(index=1, color="orange", value=75.0),
+                       Threshold(index=2, color="red", value=90.0),
+                   ],
+                   format=UNITS.PERCENT_FORMAT,
+                   thresholdMarkers=False,
+                   repeat=Repeat(direction="v", variable="server"),
+               )
+           ]
+           + [
+               # one graph/serie per host to actually generate alerts
+               Graph(
+                   title=f"{host}: trigger disk usage alert",
+                   gridPos=GridPos(h=0, w=0, x=0, y=0),  # invisible
+                   targets=[target(host)],
+                   dataSource=DATA_SOURCE,
+                   nullPointMode=NULL_AS_NULL,
+                   alert=Alert(
+                       name=f"{host} disk nearly full",
+                       message=f"{host} disk nearly full",
+                       frequency="10s",
+                       gracePeriod="1m",
+                       alertConditions=[
+                           AlertCondition(
+                               target=target(host),
+                               timeRange=TimeRange("1m", "now"),
+                               evaluator=GreaterThan(90),
+                               operator=OP_AND,
+                               reducerType=RTYPE_LAST,
+                           ),
+                       ],
+                   ),
+               )
+               for host in HOSTS
+           ],
 ).auto_panel_ids()

dashboards-to-provision/services/drand.dashboard.py

@@ -29,9 +29,9 @@
             "drand: CPU usage",
             Target(target="drand.get-cpu-percentage"),
             GridPos(h=8, w=12, x=12, y=8),
-            UNITS.PERCENT_FORMAT,
+            UNITS.NO_FORMAT,
             frequency=1 * 60,
-            alert_at=1.5,
+            alert_at=10,
         ),
         simple_graph(
             "drand.c4dt.org: TTY activity",