It is common to use ELK stack (elasticsearch, logstash, kibana) for logging. Exposing logs to everyone on team and wider organisation increases the visibility into system running in production. One can see what code paths are actually executed in production. It also makes triaging production incidents easier. What about being more proactive… Get slack notifications on errors as they happen.

elasticsearch watchers

One of elasticsearch plugins is Watcher. It enables alerts and notifications based on data in elasticsearch indexes.

bin/plugin install license
bin/plugin install watcher

Adding a watcher is a matter of a PUT request.

curl -XPUT 'http://localhost:9200/_watcher/watch/log_error_watch' -d "@watcher.json"

slack notification template

This is a slack notification template that will get you going.

{
"trigger": {
"schedule": {
"interval": "10s"
}
},
"input": {
"search": {
"request": {
"indices": [
"logstash*"
],
"body": {
"query": {
"filtered": {
"query": {
"match": {
"level": "ERROR"
}
},
"filter": {
"range": {
"@timestamp": {
"from": "{{ctx.trigger.scheduled_time}}||-5m",
"to": "{{ctx.trigger.triggered_time}}"
}
}
}
}
}
}
}
}
},
"condition": {
"compare": {
"ctx.payload.hits.total": {
"gt": 0
}
}
},
"actions": {
"send_trigger": {
"webhook": {
"method": "POST",
"scheme": "https",
"port": 443,
"host": "hooks.slack.com",
"path": "/services/XXXXXX/XXXXX/XXXXXXXX",
"body": "{ \"text\": \"encountered {{ctx.payload.hits.total}} errors in past 5 minutes \"}",
"headers": {
"Content-type": "application/json"
}
}
}
}
}