[bisq-network/roles] Monitoring Operator/Maintainer (#10)

[Harry MacFinned]

# 2018.09 Report - Fee monitoring

Reminder : Python croned task monitoring the Bisq network onion nodes for the fees. On a 5 minutes basis.
The fee page is a small json page and, additionnaly to monitoring the fees, provides also simultaneously a convenient mean to monitor the connectivity of/to the nodes.

All requests have a 10 seconds timeout, ie if the onion node doesn't answer in 10 seconds, the request is abandoned. Generally nodes answer in less than 1 second.
The fees are also archived.

The monitoring is at the moment realized by writing alerts in files + sending emails.


### Timeouts
- For the period august 1st - august 31th, 16% of the time, one onion at least was in timeout (or another exception)
- For the period sept 1st - sept 26th, 14% of the time, one onion at least was in timeout (or another exception)

I plan to investigate further to distinguish more precisely the timeouts and other exceptions.
In some cases, a detected timeout or exception, is due to the concerned node having a proper issue.
In other (many ?) cases, **a detected timeout may in fact be due to relays between the monitor and the node** (either on clearnet or on Tornet).
Pricenodes maintainers can do something when their node is concerned. When the issue is due to the net, it's out of scope.

When I notice an issue, I cross the information with http://188.166.2.139:8080/status and I drop a line on the slack when the issue seems confirmed or lasts.
This happened only twice this month, but the monitor catches correctly the timeouts only since the 20th.


### Divergence between nodes
For the period september 1st-26th 2018, there were 0 divergences lasting on 2 consecutives 5 minutes checks.
There were less than 100 divergences catched on a one check, but none lasting.
The catched divergences correspond to fee transitions, where the transition is not achieved for all nodes.


### Fees statistics
min = 10 satoshis/byte, max = 46 satoshis/byte,

satoshis/byte | occurences
--------------|---------
10            | 88%
12-20         |  9%
22-30         |  1.7%
32-40         | occured on 17 checks
46            | occured on 14 checks

Max fees weren't too high this month, and were very rare. So, nobody should have been hurted by the fees this month.

(This month, I added also recording + alerts for high fees detection.)


### Improvements
- Some alerts could be automaticaly directed to the slack.
- the timeouts can be better analyzed
- ...


NB : This monitoring is a subtask (but independant) of the pricenode monitoring task. I'll write a little report for the prices monitoring also.


-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/bisq-network/roles/issues/10#issuecomment-425174367
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.bisq.network/pipermail/bisq-github/attachments/20180927/1f9d5a49/attachment-0001.html>