Hi everyone,
I’m pretty new to Influxdb, searched a while to find some optimizations for Cisco Network Telemetry (MDT) with Influx, but couldn’t create a good config for my Project.
Hopefully some of you could give me some hints to optimize my setup.
I couldn’t upload Files, so I uploaded them to drive and linked it in here.
What I’m trying to do:
I’m working on an concept to store our Cisco Network Telemetry Data into influxdb.
The setup looks like this:
https://drive.google.com/file/d/195vcCRGYiZG6BzwVEgjkuQPTGIEuXDkJ/view?usp=sharing
- A new message comes in
- The loadbalancer sends it to one of the two Telegraf Servers which are running the “cisco_telemetry_mdt” input plugin (telinput1 + 2)
- The message is written to an Apache Kafka topic as “influx data format”. (telbuff1 + 2 +3)
- The next two Telegraf Servers (telpipe1 + 2) are pulling the Data from the Kafka Topic and send them to the InfluxDB Server (teldb1 (there is no teldb2))
Influxdb Server Resources:
OS: Debian 9 (VMware)
CPU: 16 @ 2,30Ghz
RAM: 32 GB
Storage: Full Flash, no local SSDs
The Problem:
The influxDB Server is running on nearly 100% if it has to insert the Telemetry Data.
Is there a way to optimize my config for better performance? This is a Test setup and in Production I’ll get 5 times or more Data from our Network Devices.
Configs:
Nearly the default after I played around with it:
InfluxDB Config
### Welcome to the InfluxDB configuration file.
# The values in this file override the default values used by the system if
# a config option is not specified. The commented out lines are the configuration
# field and the default value used. Uncommenting a line and changing the value
# will change the value used at runtime when the process is restarted.
# Once every 24 hours InfluxDB will report usage data to usage.influxdata.com
# The data includes a random ID, os, arch, version, the number of series and other
# usage data. No data from user databases is ever transmitted.
# Change this option to true to disable reporting.
# reporting-disabled = false
# Bind address to use for the RPC service for backup and restore.
bind-address = "172.31.147.41:8088"
###
### [meta]
###
### Controls the parameters for the Raft consensus group that stores metadata
### about the InfluxDB cluster.
###
[meta]
# Where the metadata/raft database is stored
dir = "/var/lib/influxdb/meta"
# Automatically create a default retention policy when creating a database.
# retention-autocreate = true
# If log messages are printed for the meta service
# logging-enabled = true
###
### [data]
###
### Controls where the actual shard data for InfluxDB lives and how it is
### flushed from the WAL. "dir" may need to be changed to a suitable place
### for your system, but the WAL settings are an advanced configuration. The
### defaults should work for most systems.
###
[data]
# The directory where the TSM storage engine stores TSM files.
dir = "/var/lib/influxdb/data"
# The directory where the TSM storage engine stores WAL files.
wal-dir = "/var/lib/influxdb/wal"
# The amount of time that a write will wait before fsyncing. A duration
# greater than 0 can be used to batch up multiple fsync calls. This is useful for slower
# disks or when WAL write contention is seen. A value of 0s fsyncs every write to the WAL.
# Values in the range of 0-100ms are recommended for non-SSD disks.
# wal-fsync-delay = "0s"
# The type of shard index to use for new shards. The default is an in-memory index that is
# recreated at startup. A value of "tsi1" will use a disk based index that supports higher
# cardinality datasets.
# index-version = "tsi1"
# Trace logging provides more verbose output around the tsm engine. Turning
# this on can provide more useful output for debugging tsm engine issues.
# trace-logging-enabled = false
# Whether queries should be logged before execution. Very useful for troubleshooting, but will
# log any sensitive data contained within a query.
# query-log-enabled = true
# Validates incoming writes to ensure keys only have valid unicode characters.
# This setting will incur a small overhead because every key must be checked.
# validate-keys = false
# Settings for the TSM engine
# CacheMaxMemorySize is the maximum size a shard's cache can
# reach before it starts rejecting writes.
# Valid size suffixes are k, m, or g (case insensitive, 1024 = 1k).
# Values without a size suffix are in bytes.
# cache-max-memory-size = "1g"
# CacheSnapshotMemorySize is the size at which the engine will
# snapshot the cache and write it to a TSM file, freeing up memory
# Valid size suffixes are k, m, or g (case insensitive, 1024 = 1k).
# Values without a size suffix are in bytes.
# cache-snapshot-memory-size = "25m"
# CacheSnapshotWriteColdDuration is the length of time at
# which the engine will snapshot the cache and write it to
# a new TSM file if the shard hasn't received writes or deletes
# cache-snapshot-write-cold-duration = "10m"
# CompactFullWriteColdDuration is the duration at which the engine
# will compact all TSM files in a shard if it hasn't received a
# write or delete
# compact-full-write-cold-duration = "4h"
# The maximum number of concurrent full and level compactions that can run at one time. A
# value of 0 results in 50% of runtime.GOMAXPROCS(0) used at runtime. Any number greater
# than 0 limits compactions to that value. This setting does not apply
# to cache snapshotting.
# max-concurrent-compactions = 0
# CompactThroughput is the rate limit in bytes per second that we
# will allow TSM compactions to write to disk. Note that short bursts are allowed
# to happen at a possibly larger value, set by CompactThroughputBurst
# compact-throughput = "48m"
# CompactThroughputBurst is the rate limit in bytes per second that we
# will allow TSM compactions to write to disk.
# compact-throughput-burst = "48m"
# The threshold, in bytes, when an index write-ahead log file will compact
# into an index file. Lower sizes will cause log files to be compacted more
# quickly and result in lower heap usage at the expense of write throughput.
# Higher sizes will be compacted less frequently, store more series in-memory,
# and provide higher write throughput.
# Valid size suffixes are k, m, or g (case insensitive, 1024 = 1k).
# Values without a size suffix are in bytes.
# max-index-log-file-size = "1m"
# The maximum series allowed per database before writes are dropped. This limit can prevent
# high cardinality issues at the database level. This limit can be disabled by setting it to
# 0.
# max-series-per-database = 1000000
# The maximum number of tag values per tag that are allowed before writes are dropped. This limit
# can prevent high cardinality tag values from being written to a measurement. This limit can be
# disabled by setting it to 0.
max-values-per-tag = 0
# If true, then the mmap advise value MADV_WILLNEED will be provided to the kernel with respect to
# TSM files. This setting has been found to be problematic on some kernels, and defaults to off.
# It might help users who have slow disks in some cases.
# tsm-use-madv-willneed = false
###
### [coordinator]
###
### Controls the clustering service configuration.
###
[coordinator]
# The default time a write request will wait until a "timeout" error is returned to the caller.
# write-timeout = "10s"
# The maximum number of concurrent queries allowed to be executing at one time. If a query is
# executed and exceeds this limit, an error is returned to the caller. This limit can be disabled
# by setting it to 0.
# max-concurrent-queries = 0
# The maximum time a query will is allowed to execute before being killed by the system. This limit
# can help prevent run away queries. Setting the value to 0 disables the limit.
# query-timeout = "0s"
# The time threshold when a query will be logged as a slow query. This limit can be set to help
# discover slow or resource intensive queries. Setting the value to 0 disables the slow query logging.
# log-queries-after = "0s"
# The maximum number of points a SELECT can process. A value of 0 will make
# the maximum point count unlimited. This will only be checked every second so queries will not
# be aborted immediately when hitting the limit.
# max-select-point = 0
# The maximum number of series a SELECT can run. A value of 0 will make the maximum series
# count unlimited.
# max-select-series = 0
# The maxium number of group by time bucket a SELECT can create. A value of zero will max the maximum
# number of buckets unlimited.
# max-select-buckets = 0
###
### [retention]
###
### Controls the enforcement of retention policies for evicting old data.
###
[retention]
# Determines whether retention policy enforcement enabled.
# enabled = true
# The interval of time when retention policy enforcement checks run.
# check-interval = "30m"
###
### [shard-precreation]
###
### Controls the precreation of shards, so they are available before data arrives.
### Only shards that, after creation, will have both a start- and end-time in the
### future, will ever be created. Shards are never precreated that would be wholly
### or partially in the past.
[shard-precreation]
# Determines whether shard pre-creation service is enabled.
# enabled = true
# The interval of time when the check to pre-create new shards runs.
# check-interval = "10m"
# The default period ahead of the endtime of a shard group that its successor
# group is created.
# advance-period = "30m"
###
### Controls the system self-monitoring, statistics and diagnostics.
###
### The internal database for monitoring data is created automatically if
### if it does not already exist. The target retention within this database
### is called 'monitor' and is also created with a retention period of 7 days
### and a replication factor of 1, if it does not exist. In all cases the
### this retention policy is configured as the default for the database.
[monitor]
# Whether to record statistics internally.
# store-enabled = true
# The destination database for recorded statistics
# store-database = "_internal"
# The interval at which to record statistics
# store-interval = "10s"
###
### [http]
###
### Controls how the HTTP endpoints are configured. These are the primary
### mechanism for getting data into and out of InfluxDB.
###
[http]
# Determines whether HTTP endpoint is enabled.
# enabled = true
# Determines whether the Flux query endpoint is enabled.
# flux-enabled = false
# The bind address used by the HTTP service.
# bind-address = ":8086"
# Determines whether user authentication is enabled over HTTP/HTTPS.
# auth-enabled = false
# The default realm sent back when issuing a basic auth challenge.
# realm = "InfluxDB"
# Determines whether HTTP request logging is enabled.
# log-enabled = true
# Determines whether the HTTP write request logs should be suppressed when the log is enabled.
# suppress-write-log = false
# When HTTP request logging is enabled, this option specifies the path where
# log entries should be written. If unspecified, the default is to write to stderr, which
# intermingles HTTP logs with internal InfluxDB logging.
#
# If influxd is unable to access the specified path, it will log an error and fall back to writing
# the request log to stderr.
# access-log-path = ""
# Filters which requests should be logged. Each filter is of the pattern NNN, NNX, or NXX where N is
# a number and X is a wildcard for any number. To filter all 5xx responses, use the string 5xx.
# If multiple filters are used, then only one has to match. The default is to have no filters which
# will cause every request to be printed.
# access-log-status-filters = []
# Determines whether detailed write logging is enabled.
# write-tracing = false
# Determines whether the pprof endpoint is enabled. This endpoint is used for
# troubleshooting and monitoring.
# pprof-enabled = true
# Enables a pprof endpoint that binds to localhost:6060 immediately on startup.
# This is only needed to debug startup issues.
# debug-pprof-enabled = false
# Determines whether HTTPS is enabled.
# https-enabled = false
# The SSL certificate to use when HTTPS is enabled.
# https-certificate = "/etc/ssl/influxdb.pem"
# Use a separate private key location.
# https-private-key = ""
# The JWT auth shared secret to validate requests using JSON web tokens.
# shared-secret = ""
# The default chunk size for result sets that should be chunked.
# max-row-limit = 0
# The maximum number of HTTP connections that may be open at once. New connections that
# would exceed this limit are dropped. Setting this value to 0 disables the limit.
# max-connection-limit = 0
# Enable http service over unix domain socket
# unix-socket-enabled = false
# The path of the unix domain socket.
# bind-socket = "/var/run/influxdb.sock"
# The maximum size of a client request body, in bytes. Setting this value to 0 disables the limit.
# max-body-size = 25000000
# The maximum number of writes processed concurrently.
# Setting this to 0 disables the limit.
# max-concurrent-write-limit = 0
# The maximum number of writes queued for processing.
# Setting this to 0 disables the limit.
# max-enqueued-write-limit = 0
# The maximum duration for a write to wait in the queue to be processed.
# Setting this to 0 or setting max-concurrent-write-limit to 0 disables the limit.
# enqueued-write-timeout = 0
###
### [logging]
###
### Controls how the logger emits logs to the output.
###
[logging]
# Determines which log encoder to use for logs. Available options
# are auto, logfmt, and json. auto will use a more a more user-friendly
# output format if the output terminal is a TTY, but the format is not as
# easily machine-readable. When the output is a non-TTY, auto will use
# logfmt.
# format = "auto"
# Determines which level of logs will be emitted. The available levels
# are error, warn, info, and debug. Logs that are equal to or above the
# specified level will be emitted.
# level = "info"
# Suppresses the logo output that is printed when the program is started.
# The logo is always suppressed if STDOUT is not a TTY.
# suppress-logo = false
###
### [subscriber]
###
### Controls the subscriptions, which can be used to fork a copy of all data
### received by the InfluxDB host.
###
[subscriber]
# Determines whether the subscriber service is enabled.
# enabled = true
# The default timeout for HTTP writes to subscribers.
# http-timeout = "30s"
# Allows insecure HTTPS connections to subscribers. This is useful when testing with self-
# signed certificates.
# insecure-skip-verify = false
# The path to the PEM encoded CA certs file. If the empty string, the default system certs will be used
# ca-certs = ""
# The number of writer goroutines processing the write channel.
# write-concurrency = 40
# The number of in-flight writes buffered in the write channel.
# write-buffer-size = 1000
###
### [[graphite]]
###
### Controls one or many listeners for Graphite data.
###
[[graphite]]
# Determines whether the graphite endpoint is enabled.
# enabled = false
# database = "graphite"
# retention-policy = ""
# bind-address = ":2003"
# protocol = "tcp"
# consistency-level = "one"
# These next lines control how batching works. You should have this enabled
# otherwise you could get dropped metrics or poor performance. Batching
# will buffer points in memory if you have many coming in.
# Flush if this many points get buffered
batch-size = 5000
# number of batches that may be pending in memory
# batch-pending = 10
# Flush at least this often even if we haven't hit buffer limit
# batch-timeout = "1s"
# UDP Read buffer size, 0 means OS default. UDP listener will fail if set above OS max.
# udp-read-buffer = 0
### This string joins multiple matching 'measurement' values providing more control over the final measurement name.
# separator = "."
### Default tags that will be added to all metrics. These can be overridden at the template level
### or by tags extracted from metric
# tags = ["region=us-east", "zone=1c"]
### Each template line requires a template pattern. It can have an optional
### filter before the template and separated by spaces. It can also have optional extra
### tags following the template. Multiple tags should be separated by commas and no spaces
### similar to the line protocol format. There can be only one default template.
# templates = [
# "*.app env.service.resource.measurement",
# # Default template
# "server.*",
# ]
###
### [collectd]
###
### Controls one or many listeners for collectd data.
###
[[collectd]]
# enabled = false
# bind-address = ":25826"
# database = "collectd"
# retention-policy = ""
#
# The collectd service supports either scanning a directory for multiple types
# db files, or specifying a single db file.
# typesdb = "/usr/local/share/collectd"
#
# security-level = "none"
# auth-file = "/etc/collectd/auth_file"
# These next lines control how batching works. You should have this enabled
# otherwise you could get dropped metrics or poor performance. Batching
# will buffer points in memory if you have many coming in.
# Flush if this many points get buffered
# batch-size = 5000
# Number of batches that may be pending in memory
# batch-pending = 10
# Flush at least this often even if we haven't hit buffer limit
# batch-timeout = "10s"
# UDP Read buffer size, 0 means OS default. UDP listener will fail if set above OS max.
# read-buffer = 0
# Multi-value plugins can be handled two ways.
# "split" will parse and store the multi-value plugin data into separate measurements
# "join" will parse and store the multi-value plugin as a single multi-value measurement.
# "split" is the default behavior for backward compatability with previous versions of influxdb.
# parse-multivalue-plugin = "split"
###
### [opentsdb]
###
### Controls one or many listeners for OpenTSDB data.
###
[[opentsdb]]
# enabled = false
# bind-address = ":4242"
# database = "opentsdb"
# retention-policy = ""
# consistency-level = "one"
# tls-enabled = false
# certificate= "/etc/ssl/influxdb.pem"
# Log an error for every malformed point.
# log-point-errors = true
# These next lines control how batching works. You should have this enabled
# otherwise you could get dropped metrics or poor performance. Only points
# metrics received over the telnet protocol undergo batching.
# Flush if this many points get buffered
# batch-size = 1000
# Number of batches that may be pending in memory
# batch-pending = 5
# Flush at least this often even if we haven't hit buffer limit
# batch-timeout = "1s"
###
### [[udp]]
###
### Controls the listeners for InfluxDB line protocol data via UDP.
###
[[udp]]
# enabled = false
# bind-address = ":8089"
# database = "udp"
# retention-policy = ""
# InfluxDB precision for timestamps on received points ("" or "n", "u", "ms", "s", "m", "h")
# precision = ""
# These next lines control how batching works. You should have this enabled
# otherwise you could get dropped metrics or poor performance. Batching
# will buffer points in memory if you have many coming in.
# Flush if this many points get buffered
# batch-size = 5000
# Number of batches that may be pending in memory
# batch-pending = 10
# Will flush at least this often even if we haven't hit buffer limit
# batch-timeout = "1s"
# UDP Read buffer size, 0 means OS default. UDP listener will fail if set above OS max.
# read-buffer = 0
###
### [continuous_queries]
###
### Controls how continuous queries are run within InfluxDB.
###
[continuous_queries]
# Determines whether the continuous query service is enabled.
# enabled = true
# Controls whether queries are logged when executed by the CQ service.
# log-enabled = true
# Controls whether queries are logged to the self-monitoring data store.
# query-stats-enabled = false
# interval for how often continuous queries will be checked if they need to run
# run-interval = "1s"
###
### [tls]
###
### Global configuration settings for TLS in InfluxDB.
###
[tls]
# Determines the available set of cipher suites. See https://golang.org/pkg/crypto/tls/#pkg-constants
# for a list of available ciphers, which depends on the version of Go (use the query
# SHOW DIAGNOSTICS to see the version of Go used to build InfluxDB). If not specified, uses
# the default settings from Go's crypto/tls package.
# ciphers = [
# "TLS_ECDHE_ECDSA_WITH_CHACHA20_POLY1305",
# "TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256",
# ]
# Minimum version of the tls protocol that will be negotiated. If not specified, uses the
# default settings from Go's crypto/tls package.
# min-version = "tls1.2"
# Maximum version of the tls protocol that will be negotiated. If not specified, uses the
# default settings from Go's crypto/tls package.
# max-version = "tls1.2"
Influx 'SHOW DIAGNOSTICS'
> show diagnostics
name: build
Branch Build Time Commit Version
------ ---------- ------ -------
1.7 f8fdf652f348fc9980997fe1c972e2b79ddd13b0 1.7.7
name: config
bind-address reporting-disabled
------------ ------------------
172.31.147.41:8088 false
name: config-coordinator
log-queries-after max-concurrent-queries max-select-buckets max-select-point max-select-series query-timeout write-timeout
----------------- ---------------------- ------------------ ---------------- ----------------- ------------- -------------
0s 0 0 0 0 0s 10s
name: config-cqs
enabled query-stats-enabled run-interval
------- ------------------- ------------
true false 1s
name: config-data
cache-max-memory-size cache-snapshot-memory-size cache-snapshot-write-cold-duration compact-full-write-cold-duration dir max-concurrent-compactions max-index-log-file-size max-series-per-database max-values-per-tag series-id-set-cache-size wal-dir wal-fsync-delay
--------------------- -------------------------- ---------------------------------- -------------------------------- --- -------------------------- ----------------------- ----------------------- ------------------ ------------------------ ------- ---------------
1073741824 26214400 10m0s 4h0m0s /var/lib/influxdb/data 0 1048576 1000000 0 100 /var/lib/influxdb/wal 0s
name: config-httpd
access-log-path bind-address enabled https-enabled max-connection-limit max-row-limit
--------------- ------------ ------- ------------- -------------------- -------------
:8086 true false 0 0
name: config-meta
dir
---
/var/lib/influxdb/meta
name: config-monitor
store-database store-enabled store-interval
-------------- ------------- --------------
_internal true 10s
name: config-precreator
advance-period check-interval enabled
-------------- -------------- -------
30m0s 10m0s true
name: config-retention
check-interval enabled
-------------- -------
30m0s true
name: config-subscriber
enabled http-timeout write-buffer-size write-concurrency
------- ------------ ----------------- -----------------
true 30s 1000 40
name: network
hostname
--------
ti-p-teldb1
name: runtime
GOARCH GOMAXPROCS GOOS version
------ ---------- ---- -------
amd64 16 linux go1.11
name: system
PID currentTime started uptime
--- ----------- ------- ------
8485 2019-08-26T13:51:15.775420134Z 2019-08-26T13:33:08.698080784Z 18m7.07733935s
influx 'SHOW STATS'
https://drive.google.com/open?id=1ptoe5e8B08MULaR-OXcAdPEdliP4pqd1
telinput config (telegraf.d/cisco_telemetry.conf)
[[inputs.cisco_telemetry_mdt]]
transport = "tcp"
service_address = ":5432"
[inputs.cisco_telemetry_mdt.tags]
tag_kafkatopic = "telemetry_raw"
[[outputs.kafka]]
brokers = ["172.31.147.30:9092","172.31.147.31:9092","172.31.147.32:9092"]
topic = "telemetry_raw"
client_id = "telpipe1"
compression_codec = 0
data_format = "influx"
tagexclude = ["kafkatopic"]
[outputs.kafka.tagpass]
tag_kafkatopic = ["telemetry_raw"]
telinput config (agent)
[agent]
interval = "10s"
round_interval = true
metric_batch_size = 1000
metric_buffer_limit = 10000
collection_jitter = "0s"
flush_interval = "10s"
flush_jitter = "0s"
precision = ""
debug = false
quiet = false
logfile = ""
hostname = ""
omit_hostname = false
tepipe config (telegraf.d/cisco_telemetry.conf)
[[inputs.kafka_consumer]]
brokers = ["172.31.147.30:9092","172.31.147.31:9092","172.31.147.32:9092"]
topics = ["telemetry_raw"]
client_id = "telpipe3"
consumer_group = "telpipe_telemetry_raw"
data_format = "influx"
[inputs.kafka_consumer.tags]
tag_kafkatopic = "telemetry_raw"
[[outputs.influxdb]]
urls = ["http://172.31.147.41:8086"]
database = "mdt"
skip_database_creation = true
timeout = "5s"
username = "telegraf"
password = "password"
tagexclude = ["kafkatopic"]
[outputs.influxdb.tagpass]
tag_kafkatopic = ["telemetry_raw"]
tepipe config (agent)
[agent]
interval = "1s"
round_interval = true
metric_batch_size = 50000
metric_buffer_limit = 100000
collection_jitter = "0s"
flush_interval = "500ms"
flush_jitter = "0s"
precision = ""
debug = true
quiet = false
logfile = ""
hostname = ""
omit_hostname = false
Some Monitoring Data from the Server:
https://drive.google.com/open?id=1pmYc1-bkRWL0q349qviKXcyPEmhhfHOB
https://drive.google.com/open?id=1elmnHWwvkzGIZUWp-1x_0gy2TJCYESS4
https://drive.google.com/open?id=1qQz2CXGpaXzW2rqzqtOo306b5zsS21b8
