Bump version

CHANGELOG.md

@@ -2,6 +2,17 @@
 
 ## Changes Since Last Release
 
+### Changed defaults / behaviours
+- ...
+
+### New Features & Functionality
+- ...
+
+## Bug Fixes
+- ...
+
+## 1.0.43 \[2023-08-18\]
+
 ### Changed defaults / behaviours
 - Move basic images from /deploy to /images 
 - ...
@@ -10,13 +21,15 @@
 - Add autocompletion for uninstall command.
 - Add validation for missing callable.
 - Add flag for disabling chaos controller.
+- Add flat on system charts for setting the inotify limits.
 - ...
 
 ## Bug Fixes
 - Remove VirtualObjects and Templates from the list of CRDs whose finalizers can be forcibly deleted (they have no finalizers).
 - Fix issues with partial and conditional deployments (Ingress, chaos, ..)
 - Fix platform chart to disable kubernetes dashboard according to the chart values.
 - Fix the way embedded files are coped to the host filesystem.
+- Remove single quote from cadvisor as it may not be supported on various backends, such as HPK.
 - ...
 
 
@@ -139,4 +152,4 @@ the renamed systems templates affect the controller.
 - Upgrade golang.org/x/net to v0.7.0 to avoid security bugs
 - Fix `kubectl-frisbee report` to evaluate dashboard variables (returns `.+` always).
 
-## 1.0.37 \[2023-02-02\]
+## 1.0.37 \[2023-02-02\]

charts/system/templates/telemetry/cadvisor/cadvisor.yml

@@ -32,6 +32,13 @@ spec:
           - |  # Multi-line str
             set -eux
 
+            {{- if .Values.telemetry.cadvisor.limits}}
+            # update the limits temporarily to avoid "inotify_init: too many open files"
+            sysctl fs.inotify.max_user_instances=8192
+            sysctl fs.inotify.max_user_watches=524288
+            sysctl -p
+            {{- end}}
+
             get_main_pid() {
               # Wait a bit to obtain the PID of the main container
               [ -f "/dev/shm/app" ] || inotifywait /dev/shm --include 'app'
@@ -40,8 +47,8 @@ spec:
             }
 
             enter_ns() {
-              # Check that main's PID corresponds to an actual process.
-              # If it doesn't, it may be already complete. In any case, it is not our job
+              # Check that PID of main container corresponds to an actual process.
+              # If it does not, it may be already complete. In any case, it is not our job
               # to find out the reason. Just exit with success and let experiment to continue.
               [ -d "/proc/${mainPID}" ] || { echo "/proc/${mainPID} does not exists.";  exit 0; }
 
@@ -52,11 +59,6 @@ spec:
               nsenter -t ${mainPID} -C -- mount -t cgroup2 none /sys/fs/cgroup
             }
 
-            # update the limits temporarily to avoid "inotify_init: too many open files"
-            sysctl fs.inotify.max_user_instances=8192
-            sysctl fs.inotify.max_user_watches=524288
-            sysctl -p
-
             get_main_pid && enter_ns
 
             # Start the monitoring

charts/system/values.yaml

@@ -9,6 +9,7 @@
 ## @param telemetry.prometheus.honorTimestamp Use the timestamps of the metrics exposed by the agent (time-drifts)
 ## @param telemetry.prometheus.queryLookbackDelta The maximum duration for retrieving metrics for considering the source as stale.
 ## @param telemetry.dataviewer.port Listening port for Dataviewer
+## @param telemetry.cadvisor.limits Set limits for inotify
 telemetry:
   grafana:
     image: grafana/grafana-oss:9.4.7
@@ -29,11 +30,15 @@ telemetry:
     port: 80
 
 
+  cadvisor:
+    limits: false
+
+
 ## @section Chaos
 
 ## @param chaos.network.generic.source A list of comma separated services to apply the fault
 ## @param chaos.network.generic.duration The duration of the fault
-## @param chaos.network.partition.partial.dst A list of comma seperated services that will be partitioned from the source servies.
+## @param chaos.network.partition.partial.dst A list of comma seperated services that will be partitioned from the source services.
 ## @param chaos.network.partition.partial.direction The direction of the network partition fault
 ## @param chaos.network.duplicate.duplicate  Percent of Duplicate packets
 ## @param chaos.network.duplicate.correlation  Affinity to last packet. Emulates packet burst duplicates.

examples/apps/mpi-jobs/templates/launcher.yml → examples/apps/opengadget3/templates/launcher.yml

@@ -16,13 +16,14 @@ spec:
       - name: ssh-volume
         configMap:
           name: ssh-config
+      # declare mpi volume
+      - name: mpi-volume
+        emptyDir: {}
       # declare dataset volume
       - name: dataset-volume
         hostPath:
           path: {{.Values.datasetPath}}
-      # declare mpi volume
-      - name: mpi-volume
-        emptyDir: { }
+    # Populate the hostfile with workers.
     initContainers:
       - name: populate-workers
         image: busybox

examples/apps/mpi-jobs/templates/worker.yml → examples/apps/opengadget3/templates/worker.yml

@@ -17,7 +17,6 @@ spec:
       - name: dataset-volume
         hostPath:
           path: {{.Values.datasetPath}}
-
     containers:
       - image: icsforth/opengadget3:boxes_32
         # imagePullPolicy: Always