Merge pull request #97 from delphix/dlpx/pr/aleksei-bokov/f1b9eb87-bb26-47c1-8e00-0c1545d89708

APIGW-10925 Extend support bundle data with historical docker stats

Author: aleksei-bokov

.github/workflows/flake8.yml

@@ -10,7 +10,7 @@ jobs:
     - name: Set up Python
       uses: actions/setup-python@v1
       with:
-        python-version: 3.6
+        python-version: "3.10"
     - name: flake8
       run: |
         python -m pip install --upgrade pip

cmd/nfs_threads.py

@@ -153,7 +153,7 @@ def print_line(interval):
     if e1 or e2 or e3:
         return
 
-    while(not sleep(interval)):
+    while not sleep(interval):
         nextCpuTime, e1 = cpu_time(pids)
         nextPackets, nextEnqueued, nextWoken, e2 = pool_stats()
         nextRB, nextWB, nextRO, nextWO, nextMeta, e3 = nfs_stats()

debian/control

@@ -13,6 +13,6 @@ Standards-Version: 4.1.2
 
 Package: performance-diagnostics
 Architecture: any
-Depends: python3-bcc, python3-minimal, python3-psutil, telegraf
+Depends: python3-bcc, python3-minimal, python3-psutil, telegraf, docker.io
 Description: eBPF-based Performance Diagnostic Tools
   A collection of eBPF-based tools for diagnosing performance issues.

debian/postinst

@@ -0,0 +1,27 @@
+#!/bin/bash
+
+USER="telegraf"
+GROUP="docker"
+
+# Ensure user and group exist
+if ! id "$USER" &>/dev/null; then
+	echo "Error: User '$USER' does not exist" >&2
+	exit 1
+fi
+
+if ! getent group "$GROUP" &>/dev/null; then
+	echo "Error: Group '$GROUP' does not exist" >&2
+	exit 1
+fi
+
+# Add user to the group if not already a member
+if ! groups "$USER" | grep -q "\b$GROUP\b"; then
+	if usermod -a -G "$GROUP" "$USER"; then
+		echo "User '$USER' added to group '$GROUP'."
+	else
+		echo "Error: Failed to add user '$USER' to group '$GROUP'" >&2
+		exit 1
+	fi
+fi
+
+exit 0

debian/prerm

@@ -0,0 +1,27 @@
+#!/bin/bash
+
+USER="telegraf"
+GROUP="docker"
+
+# Ensure user and group exist
+if ! id "$USER" &>/dev/null; then
+	echo "Error: User '$USER' does not exist" >&2
+	exit 1
+fi
+
+if ! getent group "$GROUP" &>/dev/null; then
+	echo "Error: Group '$GROUP' does not exist" >&2
+	exit 1
+fi
+
+# Remove user from the group
+if groups "$USER" | grep -q "\b$GROUP\b"; then
+	if gpasswd -d "$USER" "$GROUP"; then
+		echo "User '$USER' removed from group '$GROUP'."
+	else
+		echo "Error: Failed to remove user '$USER' from group '$GROUP'" >&2
+		exit 1
+	fi
+fi
+
+exit 0

telegraf/delphix-telegraf-service

@@ -1,11 +1,17 @@
 #!/bin/bash
 BASE_CONFIG=/etc/telegraf/telegraf.base
 DOSE_INPUTS=/etc/telegraf/telegraf.inputs.dose
+DCT_INPUTS=/etc/telegraf/telegraf.inputs.dct
 PLAYBOOK_INPUTS=/etc/telegraf/telegraf.inputs.playbook
 PLAYBOOK_FLAG=/etc/telegraf/PLAYBOOK_ENABLED
 TELEGRAF_CONFIG=/etc/telegraf/telegraf.conf
 
 
+function engine_is_dct() {
+	/usr/bin/get-appliance-variant | grep -q "dct" >/dev/null
+	[[ "$?" == "0" ]]
+}
+
 function engine_is_object_based() {
 	zdb -C | grep "type: 'object_store'" >/dev/null
 	[[ "$?" == "0" ]]
@@ -23,6 +29,12 @@ if engine_is_object_based; then
 	else
 		cat $DOSE_INPUTS $BASE_CONFIG > $TELEGRAF_CONFIG
 	fi
+elif engine_is_dct; then
+	if playbook_is_enabled; then
+		cat $PLAYBOOK_INPUTS $DCT_INPUTS $BASE_CONFIG > $TELEGRAF_CONFIG
+	else
+		cat $DCT_INPUTS $BASE_CONFIG > $TELEGRAF_CONFIG
+	fi
 else
 	if playbook_is_enabled; then
 		cat $PLAYBOOK_INPUTS $BASE_CONFIG > $TELEGRAF_CONFIG

telegraf/telegraf.base

@@ -18,7 +18,7 @@
   rotation_max_size = "50MB"
   rotation_max_archives = 9
   data_format = "json"
-  namedrop = ["*estat_*", "agg_*", "zfs", "zpool*", "zcache*"]
+  namedrop = ["*estat_*", "agg_*", "zfs", "zpool*", "zcache*", "docker*"]
 
 # Define output file for ZFS related metrics
 [[outputs.file]]

telegraf/telegraf.inputs.dct

@@ -0,0 +1,22 @@
+#######################  Docker/DCT services Metrics  ################################
+[[inputs.docker]]
+  endpoint = "unix:///var/run/docker.sock"
+  interval = "60s"
+  timeout = "10s"
+  namepass = [
+    "docker_container_cpu",
+    "docker_container_mem",
+    "docker_container_blkio",
+    "docker_container_status"
+  ]
+  docker_label_exclude = ["com.docker.compose.*", "resty*"]
+
+ [[outputs.file]]
+  files = ["/var/log/telegraf/metrics_docker.json"]
+  rotation_max_size = "30MB"
+  rotation_max_archives = 5
+  data_format = "json"
+  namepass = ["docker*"]
+
+####################### End of Docker/DCT services Metrics #######################
+