DNSCrypt Proxy Monitoring Dashboard inconveniences

[ghost]

1. Order of entries in Query Type changes all the time, even without data changes.
2. Top Domains seems to show a random sample of the domains with a given count, therefore the displayed domains change all the time, even without data changes.
3. Top Domains column widths change in rare cases.
4. Dashboard automatically scrolls to different position in rare cases.

[jedisct1]

If it is inconvenient, keep it disabled.

[ghost]

The number of Recent Queries is a bit low. E.g. just starting the pc and opening the browser to see the dashboard might causes too many queries to have them all displayed for analysis.

Also Top Blocked Domains would be nice to have.

[jedisct1]

You can enable the query log

[ghost]

Log files stop logging when the pc gets restarted. Then dnscrypt-proxy needs to be restarted to restart the logging which i want to avoid.

The dashboard however works continously despite pc restarts. That's more useful.

[ghost]

With 2.1.12 the flickering is gone (thanks!), however auto-scrolling is still happening.

[jedisct1]

Works for me. Fix it.

[ghost]

Websocket is sporadically closed by firefox. Probably ffx considers it dead because of some implementation incompatibilities.

[ghost]

What i found out is that if a websocket is reconnected there's a short moment when the recent queries section is gone from the dashboard. Obviously this causes ffx to scroll up to still existing content.
Anyone having any ideas how to fix that?

[ghost]

Tested 91e20e4 and it seems to work. Thanks a lot!

Next one:
When the first websocket ws is ever closed, pollMetrics() starts polling and will never stop as it is unable to recognize that a new websocket newWs was successfully started.

edit:

1. In connectWebSocket() there is a comment claiming We can't update the global ws variable from here. Why wouldn't that be possible (except for having to rename the local variable ws to something else)?
2. Should the websocket be reconnected endlessly even if it's closed again and again?

[ghost]

Just saw a successful response in the recent queries with over 3*10^7 ms ~ 10h. How is that possible?