Logsend is high-performance tool for processing logs.
This like Logstash but tiny and written by Golang. Supported outputs:
Just download binary by your platform from GitHub Latest Releases and unzip, or use this installer
Install logsend to /usr/local/bin
curl -L https://logsend.io/get|sudo bash
OR to your own directory
curl -L https://logsend.io/get|bash -s /tmp
As daemon, watching file in directory:
logsend -config=config.json /logs
Using PIPE:
tail -F /logs/*.log |logsend -config=config.json
cat /logs/*.log |logsend -config=config.json
ssh user@host "tail -F /logs/*.log"|logsend -config=config.json
Or using PIPE without config.json:
tail -F /logs/*.log |logsend -influx-dbname test -influx-host 'hosta:4444' -regex='\d+'
Daemonize:
logsend -config=config.json /logs 2> error.log &
Log files | Lines per log | Result (real/user/sys) |
---|---|---|
1 | 250k | 0m3.186s 0m3.742s 0m1.672s |
5 | 250k | 0m5.641s 0m13.423s 0m1.015s |
10 | 250k | 0m8.577s 0m25.781s 0m1.341s |
config.json
{
"influxdb": {
"host": "localhost:4444",
"user": "root",
"password": "root",
"database": "logers",
"udp": true,
"send_buffer": 8
},
"groups": [
{
"mask": "test.log",
"rules": [
{
"regexp": "test string (?P<word_STRING>\\\\w+)",
"influxdb": {
"name": "test"
}
}
]
}
]
}
Log files | Lines per log | Result (real/user/sys) |
---|---|---|
1 | 250k | 0m5.722s 0m7.638s 0m3.779s |
5 | 250k | 0m9.676s 0m30.553s 0m2.410s |
10 | 250k | 0m18.385s 0m58.795s 0m2.534s |
config.json
{
"influxdb": {
"host": "localhost:4444",
"user": "root",
"password": "root",
"database": "logers",
"udp": true,
"send_buffer": 8
},
"groups": [
{
"mask": "test.log",
"rules": [
{
"regexp": "test string (?P<word_STRING>\\\\w+)",
"influxdb": {
"name": "test"
}
},
{
"regexp": "(?P<word_STRING>\\\\w+) string one",
"influxdb": {
"name": "test"
}
},
{
"regexp": "string (?P<word_STRING>\\\\w+) one",
"influxdb": {
"name": "test"
}
}
]
}
]
}
logsend -config config.json /logs
Influxdb configuration
Description:
"mask": "search.log$"
- match logs in watch-dir directory
"regexp": "metric dbcount (?P<count_INT>[0-9]+)$",
- regexp for matching line. But we want to store some values from this line and we should use named group with type association, in this case count_INT
will be perform to field count
with type int
. Also supported notation FLOAT
, STRING
.
"name": "keys_count"
- database name in Influxdb
{
"influxdb": {
"host": "host:4444",
"user": "root",
"password": "root",
"database": "logers"
},
"groups": [
{
"mask": "search.log$",
"rules": [
{
"regexp": "metric dbcount (?P<count_INT>[0-9]+)$",
"influxdb": {
"name": "keys_count"
}
}
]
}
]
}
Advanced Influxdb configuration
Description:
"udp": true,
- sending packets thought UDP(require that Influxdb supported this option)
"send_buffer": 8
- messages per one packet. But don't forget, max size of packet should be no more then 1500 Byte
Some times we need added not captured params, and for this purpose we used extra_fields
key.
"extra_fields": [
["service", "search"],
["host", "HOST"]
]
["service", "search"],
- First is column name, second it's value.
["host", "HOST"]
- HOST it's a special keyword who's returned hostname of machine.
{
"influxdb": {
"host": "host:4444",
"user": "root",
"password": "root",
"database": "logers",
"udp": true,
"send_buffer": 8
},
"groups": [
{
"mask": "search.log$",
"rules": [
{
"regexp": "metric dbcount (?P<count_INT>[0-9]+)$",
"influxdb": {
"name": "keys_count",
"extra_fields": [
["service", "search"],
["host", "HOST"]
]
}
}
]
}
]
}
NewRelic configuration
Description:
"host"
- hostname when started logsend"name"
- application name"key"
- api key for NewRelic"duration"
- seconds, how often send report to New Relic"tmpl"
- metric name
{
"newrelic": {
"host": "host1",
"key": "0000023231232231",
"name": "project1"
},
"groups": [
{
"mask": "access.log$",
"rules": [
{
"newrelic": {
"duration": 10,
"tmpl": "Component/Project1/NginxRequestsCode{{.code}}"
},
"regexp": "\\d+\\.\\d+\\.\\d+\\.\\d+ .+ \\[.+\\] \"\\w+ \\S+ \\S+\" (?P<code>\\d+)"
},
{
"newrelic": {
"duration": 10,
"tmpl": "Component/Project1/OpenPageType-{{.pageType}}"
},
"regexp": "\\d+\\.\\d+\\.\\d+\\.\\d+ .+ \\[.+\\] \"\\w+ /(?P<pageType>hotels|cities|countries)/\\S+ \\S+\" 200"
}
]
},
{
"mask": "production.log$",
"rules": [
{
"newrelic": {
"duration": 10,
"tmpl": "Component/Project1/RailsRequestsCode{{.code}}"
},
"regexp": "^Completed (?P<code>\\d+)"
},
{
"newrelic": {
"duration": 10,
"tmpl": "Component/Project1/RailsResponseTime{{.NAME}}"
},
"regexp": "^Completed \\d+ .+ in \\d+\\.\\d+ms \\(Views: (?P<RenderViews_FLOAT>\\d+\\.\\d+)ms .+ ActiveRecord: (?P<ActiveRecord_FLOAT>\\d+\\.\\d+)ms"
}
]
}
]
}
Statsd configuration
Description:
["search.prepare_proposal_time", "prepare_proposals"]
- set prepare_proposals
to search.prepare_proposal_time
as timing
"search.prepare_proposals"
- increment by by this metric
["search.keys_count", "prepare_proposals"]
- set non integer metrics
{
"regexp": "PreapreProposals took (?P<prepare_proposals_DurationToMillisecond>.+)$",
"statsd": {
"timing": [
["search.prepare_proposal_time", "prepare_proposals"]
],
"increment": [
"search.prepare_proposals"
],
"gauge": [
["search.keys_count", "prepare_proposals"]
]
}
}
Full Influx and Statsd configuration
{
"influxdb": {
"host": "hosta:4444",
"user": "root",
"password": "root",
"database": "logers",
"udp": true,
"send_buffer": 8
},
"statsd": {
"host": "hostb:8125",
"prefix": "test.",
"interval": "1s"
},
"groups": [
{
"mask": "search.log$",
"rules": [
{
"regexp": "metric dbcount (?P<count_INT>[0-9]+)$",
"influxdb": {
"name": "g_keys_count",
"extra_fields": [
["service", "search"],
["host", "HOST"]
]
},
"statsd": {
"gauge": [
["search.keys_count", "count"]
]
}
},
{
"regexp": "PreapreProposals took (?P<prepare_proposals_DurationToMillisecond>.+)$",
"influxdb": {
"name": "g_benchmark",
"extra_fields": [
["service", "search"],
["host", "HOST"]
]
},
"statsd": {
"timing": [
["search.prepare_proposal_time", "prepare_proposals"]
],
"increment": [
"search.prepare_proposals"
]
}
},
{
"regexp": "Completed (?P<code_INT>\\d+) .+ in (?P<tm_DurationToMillisecond>.+)$",
"influxdb": {
"name": "g_responses",
"extra_fields": [
["service", "search"],
["host", "HOST"]
]
},
"statsd": {
"increment": [
"search.requests"
]
}
}
]
},
{
"mask": "map.log$",
"rules": [
{
"regexp": "metric dbcount (?P<count_INT>[0-9]+)$",
"influxdb": {
"name": "g_keys_count",
"extra_fields": [
["service", "map"],
["host", "HOST"]
]
}
},
{
"regexp": "PreapreProposals took (?P<prepare_proposals_DurationToMillisecond>.+)$",
"influxdb": {
"name": "g_benchmark",
"extra_fields": [
["service", "map"],
["host", "HOST"]
]
}
},
{
"regexp": "Completed (?P<code_INT>\\d+) .+ in (?P<tm_DurationToMillisecond>.+)$",
"influxdb": {
"name": "g_responses",
"extra_fields": [
["service", "map"],
["host", "HOST"]
]
}
}
]
}
]
}
config.json:
{
"influxdb": {
"host": "influxdbhost:8086",
"user": "root",
"password": "root",
"database": "logers",
"send_buffer": 12
},
"groups": [
{
"mask": "some.[0-9]+.log$",
"rules": [
{
"regexp": "\\[W .+ chain_processor.+\\] \\[.+\\] (?P<gate_STRING>\\w+_gate) -> (?P<exec_time_FLOAT>\\d+.\\d+)",
"influxdb": {
"name": "gate_response",
"extra_fields": [
["host", "HOST"]
]
}
},
{
"regexp": "^\\[I .+ web.+\\] (?P<code_INT>\\d+) \\w+ (?P<name_STRING>\/\\S+)(?:\\?.+) \\(.+\\) (?P<resp_time_FLOAT>\\d+.\\d+)ms",
"influxdb": {
"name": "adaptors",
"extra_fields": [
["host", "HOST"]
]
}
}
]
}
]
}
another examples how to run:
only one rule:
tail -F some.log| logsend -influxdb-host "influxdbhost:8086"
-influxdb-user root -influxdb-password root
-influxdb-database logers -influxdb-send_buffer 12
-influxdb-extra_fields 'host,HOST'
-regex '\[W .+ chain_processor.+\] \[.+\] (?P<gate_STRING>\w+_gate) -> (?P<exec_time_FLOAT>\d+.\d+)'
using previous config file:
tail -F some.log| logsend -config config.json
config.json:
{
"statsd": {
"host": "statsdhost:8125",
"prefix": "test",
"interval": "1s"
},
"groups": [
{
"mask": "some.[0-9]+.log$",
"rules": [
{
"regexp": "\\[W .+ chain_processor.+\\] \\[.+\\] (?P<gate_STRING>\\w+_gate) -> (?P<exec_time_FLOAT>\\d+.\\d+)",
"statsd": {
"timing": [
["gate_exec_time", "exec_time"]
]
}
},
{
"regexp": "^\\[I .+ web.+\\] (?P<code_INT>\\d+) \\w+ (?P<name_STRING>\/\\S+)(?:\\?.+) \\(.+\\) (?P<resp_time_FLOAT>\\d+.\\d+)ms",
"statsd": {
"increment": [
"adaptors_exec_count"
]
}
}
]
}
]
}
another examples how to run:
only one rule:
tail -F some.log| logsend -statsd-host "statsdhost:8125"
-statsd-host "statsdhost:8125" -statsd-prefix test
-statsd-interval 1s -statsd-timing 'gate_exec_time,exec_time'
-regex '\[W .+ chain_processor.+\] \[.+\] (?P<gate_STRING>\w+_gate) -> (?P<exec_time_FLOAT>\d+.\d+)'
using previous config file:
tail -F some.log| logsend -config config.json
config.json:
{
"mysql": {
"host": "root:toor@/test1?timeout=30s&strict=true"
},
"groups": [
{
"mask": "some.[0-9]+.log$",
"rules": [
{
"regexp": "\\[W .+ chain_processor.+\\] \\[.+\\] (?P<gate_STRING>\\w+_gate) -> (?P<exec_time_FLOAT>\\d+.\\d+)",
"mysql": {
"query": [
"insert into test1(teststring, testfloat) values('{{.gate}}', {{.exec_time}});",
"insert into test2(teststring, testfloat) values('{{.gate}}', {{.exec_time}});"
]
}
},
{
"regexp": "^\\[I .+ web.+\\] (?P<code_INT>\\d+) \\w+ (?P<name_STRING>\/\\S+)(?:\\?.+) \\(.+\\) (?P<resp_time_FLOAT>\\d+.\\d+)ms",
"mysql": {
"query": [
"insert into test1(teststring, testfloat) values('a', 22);"
]
}
}
]
}
]
}
another examples how to run:
only one rule:
tail -F some.log| logsend -mysql-host "root:toor@/test1?timeout=30s&strict=true"
-mysql-query "insert into test1(teststring, testfloat) values('{{.gate}}', {{.exec_time}});"
-regex '\[W .+ chain_processor.+\] \[.+\] (?P<gate_STRING>\w+_gate) -> (?P<exec_time_FLOAT>\d+.\d+)'
using previous config file:
tail -F some.log| logsend -config config.json
- use flag
-debug
for more info - use flag
-dry-run
for processing logs but not send to destination - use flag
-read-whole-log
for reading whole log file and continue reading - use flag
-read-once
better for use with -read-whole-log, just read whole log and exit
check process logsend pidfile /home/user/logsend/logsend.pid
start program = "/bin/bash -c '/usr/local/bin/logsend -config /home/user/rails_app/current/config/logsend.json -log=/home/user/logsend/logsend.log /home/user/rails_app/shared/log & echo $! > /home/user/logsend/logsend.pid'"
as uid user and gid user
stop program = "/bin/bash -c '/bin/kill `cat /home/user/logsend/logsend.pid`'"
as uid user and gid user with timeout 5 seconds
if totalmemory is greater than 64 MB for 5 cycles then restart
if totalcpu is greater than 60% for 5 cycles then restart
if children is greater than 64 for 5 cycles then restart
check file logsend_restart path /home/user/rails_app/current/config/logsend.json
if changed timestamp then exec "/bin/bash -c '/bin/kill `cat /home/user/logsend/logsend.pid`'"
Logsend is released under the MIT License.