Skip to content

Latest commit

 

History

History
134 lines (86 loc) · 4.43 KB

INFLUXDB.md

File metadata and controls

134 lines (86 loc) · 4.43 KB
Raw

InfluxDB Statistics

So you enabled InfluxDB statistics. Hopefully you run version 1.0+. If not, stop now and upgrade.

Here you can read what data are collected and what queries you can run.

See the data

You can use InfluxDB admin panel or install Grafana

Admin panel

When influxd or influx service is running, open localhost:8083

Grafana

By default Grafana runs on port 3000. In order to solve the conflict with trustroots application you need to change the default port in configuration.

  • locate your grafana configuration file
  • find, uncomment and change http_port value according to your preference
  • save the changes and start/restart the grafana server (i.e. systemctl start grafana, it is system specific)
  • run the client on localhost:(your-value)
  • we'll provide a dashboard which you can import and use with the data.

Metrics

Read more about how to query InfluxDB. On the website look for more chapters in Contents.

Messages

These are the data stored for every message in InfluxDB

measurement: 'messageSent'
time

fields: {
  messageId: string,
  idFrom: string,
  idTo: string,
  messageLength: number, // counting only plainText of the message content
  replyTime: number [ms], // field time between the first message and the first reply
                          // available only for "position"='firstReply', otherwise -1
}

tags: {
  messageLengthType: string, // (short|long)
  position: string, // (first|firstReply|other) position in the thread
}

Amounts

How many messages are sent weekly?
SELECT COUNT(messageLength) FROM messageSent WHERE time > now() - 100w GROUP BY time(1w) fill(0)

Reply rate

How many new threads are started? How many new threads are replied?
SELECT COUNT(messageLength) FROM messageSent WHERE time > now() - 100w AND position = 'first' GROUP BY time(1w) fill(0)
SELECT COUNT(messageLength) FROM messageSent WHERE time > now() - 100w AND position = 'firstReply' GROUP BY time(1w) fill(0)
What is the ratio at given time?

Note: Messages and replies are decoupled in the database. Here we show the ratio between amount of new threads written now and first replies replied now.

  • counting data to a new measurement messageRatio with SELECT INTO
SELECT COUNT(messageLength) AS firstMessageCount INTO messageRatio FROM messageSent WHERE time > now() - 100w position='first' GROUP BY time(1w) fill(0)
SELECT COUNT(messageLength) AS firstReplyCount INTO messageRatio FROM messageSent WHERE time > now() - 100w position='firstReply' GROUP BY time(1w) fill(0)
  • comparing the data
SELECT SUM(firstReplyCount)/SUM(firstMessageCount) AS percent FROM messageRatio WHERE time > now() - 100w GROUP BY time(1w) fill(0)
  • keeping the messageRatio updated with CONTINUOUS QUERIES
CREATE CONTINUOUS QUERY "firstMessageCountCQ" ON "trustroots" RESAMPLE FOR 2w BEGIN SELECT COUNT(messageLength) AS firstMessageCount INTO messageRatio FROM messageSent WHERE position='first' GROUP BY time(1w) END

... and similarly for firstReplyCount.

Reply time

What is the time since the first message till the first reply?
  • Average reply time may be interesting:
SELECT MEAN(replyTime) FROM messageSent WHERE time > now() - 100w AND position='firstReply' GROUP BY time(1w)
  • Median reply time will be influenced less by very late replies.
SELECT MEDIAN(replyTime) FROM messageSent WHERE time > now() - 100w AND position='firstReply' GROUP BY time(1w)
How many replies are useful (replied within 1,3,7,14 days)?
SELECT COUNT(messageLength) FROM messageSent WHERE time > now() - 100w AND position = 'firstReply' AND replyTime < 24*3600*1000 GROUP BY time(1w) fill(0)

Message length

How frequent are long and short messages?
SELECT COUNT(messageLength) FROM messageSent WHERE time > now() - 100w AND messageLengthType='short' GROUP BY time(1w)
SELECT COUNT(messageLength) FROM messageSent WHERE time > now() - 100w AND messageLengthType='long' GROUP BY time(1w)
What is the ratio?

To be continued...

You can also filter by the position in the thread. You can try to write your own query which will show average and median message length.

Users

Offers