Merge pull request #81 from scottmdlpx/master

CP-8403 Adding Telegraf-based metric collection.

Author: scottmdlpx

cmd/estat.py

@@ -89,6 +89,7 @@ def die(*args, **kwargs):
       -q/-Q     enable/disable latency histograms by size (default: off)
       -y/-Y     enable/disable the summary output (default: on)
       -t/-T     enable/disable emitting the summary total (default: on)
+      -j        set output mode to JSON
       -d LEVEL  set BCC debug level
       -e        emit the resulting eBPF script without executing it
 
@@ -111,7 +112,6 @@ def die(*args, **kwargs):
       particular the time spent allocating a block and time spent waiting for
       the write I/O to complete. If POOL is not specified, defaults to tracing
       the pool 'domain0'.
-
 """
 
 
@@ -149,6 +149,7 @@ def usage(msg):
 script_arg = None
 debug_level = 0
 dump_bpf = False
+output_mode = BCCHelper.ESTAT_PRINT_MODE
 
 
 class Args:
@@ -161,6 +162,7 @@ class Args:
 setattr(args, "latsize_hist", False)
 setattr(args, "summary", True)
 setattr(args, "total", True)
+setattr(args, "json", False)
 
 #
 # We use getopt rather than argparse because it is very difficult to get
@@ -170,7 +172,7 @@ class Args:
 # arguments.
 #
 try:
-    opts, rem_args = getopt.getopt(sys.argv[2:], "hmMa:lLzZqQyYnNtTd:e")
+    opts, rem_args = getopt.getopt(sys.argv[2:], "hmMa:lLjzZqQyYnNtTd:e")
 except getopt.GetoptError as err:
     die(err)
 
@@ -194,6 +196,7 @@ class Args:
         dump_bpf = True
     else:
         switches = {'-l': "lat_hist",
+                    '-j': "json",
                     '-z': "size_hist",
                     '-q': "latsize_hist",
                     '-y': "summary",
@@ -219,6 +222,9 @@ class Args:
 if not (args.lat_hist or args.size_hist or args.latsize_hist):
     args.lat_hist = True
 
+if args.json:
+    output_mode = BCCHelper.ANALYTICS_PRINT_MODE
+
 # Now that we are done parsing arguments, construct the text of the BPF program
 try:
     with open(base_dir + 'bpf/estat/' + program + '.c', 'r') as prog_file:
@@ -443,7 +449,7 @@ class Args:
                 probe_type + "'")
 
 if args.lat_hist or args.size_hist or args.summary:
-    helper1 = BCCHelper(b, BCCHelper.ESTAT_PRINT_MODE)
+    helper1 = BCCHelper(b, output_mode)
     helper1.add_key_type("name")
     helper1.add_key_type("axis")
 
@@ -465,23 +471,24 @@ class Args:
                                 "bytes")
 
 if args.latsize_hist:
-    helper2 = BCCHelper(b, BCCHelper.ESTAT_PRINT_MODE)
+    helper2 = BCCHelper(b, output_mode)
     helper2.add_aggregation("latsq", BCCHelper.LL_HISTOGRAM_AGGREGATION,
                             "microseconds")
     helper2.add_key_type("size")
     helper2.add_key_type("name")
     helper2.add_key_type("axis")
 
 if args.summary and args.total:
-    helper3 = BCCHelper(b, BCCHelper.ESTAT_PRINT_MODE)
+    helper3 = BCCHelper(b, output_mode)
     helper3.add_aggregation("opst", BCCHelper.COUNT_AGGREGATION, "iops(/s)")
     helper3.add_aggregation("datat", BCCHelper.SUM_AGGREGATION,
                             "throughput(k/s)")
     helper3.add_key_type("name")
 
 # Need real time;
-print("%-16s\n" % strftime("%D - %H:%M:%S %Z"))  # TODO deduplicate this line
-print(" Tracing enabled... Hit Ctrl-C to end.")
+if not args.json:
+    print("%-16s\n" % strftime("%D - %H:%M:%S %Z"))  # TODO deduplicate line
+    print(" Tracing enabled... Hit Ctrl-C to end.")
 
 # output
 if monitor:
@@ -508,7 +515,8 @@ class Args:
                 helper1.printall(clear_data)
             if args.summary and args.total:
                 helper3.printall(clear_data)
-            print("%-16s\n" % strftime("%D - %H:%M:%S %Z"))
+            if not args.json:
+                print("%-16s\n" % strftime("%D - %H:%M:%S %Z"))
         except Exception as e:
             die(e)
 else:

debian/rules

@@ -22,3 +22,6 @@ override_dh_auto_install:
 	dh_install build/cmd/* /usr/bin
 	dh_install lib/* /usr/share/performance-diagnostics/lib
 	dh_install bpf/* /usr/share/performance-diagnostics/bpf
+	dh_install telegraf/delphix-telegraf-service telegraf/perf_playbook /usr/bin
+	dh_install telegraf/delphix-telegraf.service /lib/systemd/system
+	dh_install telegraf/telegraf* telegraf/*.sh /etc/telegraf

telegraf/delphix-telegraf-service

@@ -0,0 +1,34 @@
+#!/bin/bash
+BASE_CONFIG=/etc/telegraf/telegraf.base
+DOSE_INPUTS=/etc/telegraf/telegraf.inputs.dose
+PLAYBOOK_INPUTS=/etc/telegraf/telegraf.inputs.playbook
+PLAYBOOK_FLAG=/etc/telegraf/PLAYBOOK_ENABLED
+TELEGRAF_CONFIG=/etc/telegraf/telegraf.conf
+
+
+function engine_is_object_based() {
+	zdb -C | grep "type: 'object_store'" >/dev/null
+	[[ "$?" == "0" ]]
+}
+
+function playbook_is_enabled() {
+	[[ -f $PLAYBOOK_FLAG ]]
+}
+
+rm -f $TELEGRAF_CONFIG
+
+if engine_is_object_based; then
+	if playbook_is_enabled; then
+		cat $PLAYBOOK_INPUTS $DOSE_INPUTS $BASE_CONFIG > $TELEGRAF_CONFIG
+	else
+		cat $DOSE_INPUTS $BASE_CONFIG > $TELEGRAF_CONFIG
+	fi
+else
+	if playbook_is_enabled; then
+		cat $PLAYBOOK_INPUTS $BASE_CONFIG > $TELEGRAF_CONFIG
+	else
+		cat $BASE_CONFIG > $TELEGRAF_CONFIG
+	fi
+fi
+
+/usr/bin/telegraf -config $TELEGRAF_CONFIG

telegraf/delphix-telegraf.service

@@ -0,0 +1,18 @@
+[Unit]
+Description=Delphix Telegraf Metric Collection Agent
+Documentation=https://github.com/influxdata/telegraf
+PartOf=delphix.target
+After=delphix-platform.service
+PartOf=delphix-platform.service
+
+[Service]
+EnvironmentFile=-/etc/default/telegraf
+User=root
+ExecStart=/usr/bin/delphix-telegraf-service
+ExecReload=/bin/kill -HUP $MAINPID
+Restart=on-failure
+RestartForceExitStatus=SIGPIPE
+KillMode=control-group
+
+[Install]
+WantedBy=delphix.target

telegraf/nfs-threads.sh

@@ -0,0 +1,3 @@
+#!/bin/sh
+nfs_threads | egrep --line-buffered -v "thr"
+

telegraf/perf_playbook

@@ -0,0 +1,53 @@
+#!/bin/bash
+#
+# Copyright (c) 2021 by Delphix. All rights reserved.
+#
+# Script that enables and, disables the Performance Playbook configuration for 
+# metric collection by Telegraf
+#
+
+PLAYBOOK_FLAG=/etc/telegraf/PLAYBOOK_ENABLED
+
+#
+# Make sure this can only be run as root.
+#
+function die() {
+	echo -e "$(date +%T:%N:%z): $(basename $0): $*" >&2
+	exit 1
+}
+
+[[ $EUID -ne 0 ]] && die "must be run as root"
+
+#
+# Process command.
+#
+
+function usage() {
+	echo "$(basename $0): $*" >&2
+	echo "Usage: $(basename $0) [enable|disable]"
+	exit 2
+}
+
+function enable_playbook() {
+	date
+	echo "Enabling Performance Playbook Metrics"
+	touch $PLAYBOOK_FLAG
+	systemctl restart delphix-telegraf
+}
+
+function disable_playbook() {
+	date
+	echo "Disabling Performance Playbook Metrics"
+	rm -rf $PLAYBOOK_FLAG
+	systemctl restart delphix-telegraf
+}
+
+if [[ $# -ne 1 ]]; then
+	usage
+fi
+
+case "$1" in
+enable) enable_playbook ;;
+disable) disable_playbook ;;
+*) usage ;;
+esac

telegraf/telegraf.base

@@ -0,0 +1,131 @@
+# Telegraf Configuration
+#
+# Configuration for telegraf agent
+[agent]
+  interval = "10s"
+  round_interval = true
+  flush_interval = "10s"
+  metric_batch_size = 1000
+  metric_buffer_limit = 10000
+
+###############################################################################
+#                            OUTPUT PLUGINS                                   #
+###############################################################################
+# Define the main metric output file, excluding aggregated stats and 
+# Performance Playbook (estat) data.
+[[outputs.file]]
+  files = ["/var/log/telegraf/metrics.json"]
+  rotation_max_size = "50MB"
+  rotation_max_archives = 9
+  data_format = "json"
+  namedrop = ["*estat_*", "agg_*", "zfs", "zpool*", "zcache*"]
+
+# Define output file for ZFS related metrics
+[[outputs.file]]
+  files = ["/var/log/telegraf/metrics_zfs.json"]
+  rotation_max_size = "30MB"
+  rotation_max_archives = 5
+  data_format = "json"
+  namepass = ["zpool*", "zcache*", "zfs"]
+
+# Define output file for Performance Playbook (estat) metrics
+[[outputs.file]]
+  files = ["/var/log/telegraf/metrics_estat.json"]
+  rotation_max_size = "30MB"
+  rotation_max_archives = 5
+  data_format = "json"
+  namepass = ["*estat_*"]
+
+# Define output file for aggregate statistics
+[[outputs.file]]
+  files = ["/var/log/telegraf/metric_aggregates.json"]
+  rotation_max_size = "30MB"
+  rotation_max_archives = 5
+  data_format = "json"
+  namepass = ["agg_*"]
+
+# Enable Live Monitoring, intended for internal use:
+#[[outputs.influxdb]]
+#  urls = ["http://dbsvr.company.com:8086"]
+#  database = "live_metrics"
+#  skip_database_creation = true
+#  data_format = "influx"
+
+###############################################################################
+#                            INPUT PLUGINS                                    #
+###############################################################################
+
+# Get CPU usage
+[[inputs.cpu]]
+  percpu = true
+  totalcpu = true
+  collect_cpu_time = false
+  report_active = false
+  fieldpass = ["usage*"]
+
+# Get mount point stats
+[[inputs.disk]]
+  mount_points = ["/","/domain0"]
+
+# Get disk I/O stats
+[[inputs.diskio]]
+
+# Track stats for the current metric files
+[[inputs.filestat]]
+  files = ["/var/log/telegraf/metrics.json",
+  	   "/var/log/telegraf/metrics_estat.json",
+  	   "/var/log/telegraf/metrics_zfs.json",
+  	   "/var/log/telegraf/metric_aggregates.json"]
+
+# Get Memory stats
+[[inputs.mem]]
+
+# Get some network interface stats
+[[inputs.net]]
+  fieldpass = ["tcp*","bytes*","packets*","err*","drop*"]
+
+# Track CPU and Memory for the "delphix-mgmt" service (and children).
+[[inputs.procstat]]
+  systemd_unit = "delphix-mgmt.service"
+  include_systemd_children = true
+  namedrop = ["procstat_lookup"]
+  fieldpass = ["memory_usage", "cpu_usage", "memory_rss"]
+
+# Track CPU and Memory for the "zfs-object-agent" service (and children).
+[[inputs.procstat]]
+  systemd_unit = "zfs-object-agent.service"
+  include_systemd_children = true
+  namedrop = ["procstat_lookup"]
+  fieldpass = ["memory_usage", "cpu_usage", "memory_rss"]
+
+# Get process counts
+[[inputs.processes]]
+
+# Get swap memory usage
+[[inputs.swap]]
+
+# Get misc 'other' stats (load and uptime)
+[[inputs.system]]
+
+# ZFS kstats (arcstat, abdstat, zfetch, etc)
+[[inputs.zfs]]
+  interval = "1m"
+
+# Detailed ZFS pool metrics from "zpool_influxdb" (noisy)
+#[[inputs.exec]]
+#  commands = ["/usr/lib/x86_64-linux-gnu/zfs/zpool_influxdb"]
+#  data_format = "influx"
+
+###############################################################################
+#                       AGGREGATION PLUGINS                                   #
+###############################################################################
+# Filtered aggregate statistics
+# Calculate Min, Max, Mean, Std Deviation every hour for selected metrics:
+# 	CPU Usage (%idle)
+[[aggregators.basicstats]]
+  period = "1h"
+  drop_original = false
+  stats = ["min", "max", "mean", "stdev"]
+  name_prefix = "agg_"
+  namepass = ["cpu","disk","diskio","mem","net","processes","system","swap"]
+