claashk: unified_api_buffering

[r00t-]

I implemented a joint buffer for these two APIs, which I am currently testing in my local home environment. Would you be interested in a PR?
-- #694 (comment)

@claashk: thanks for the offer, i'll have a look

[isarrider]

@claashk also with this PR it seems to not work (with VictoriaMetrics):

[isarrider]

the only thing that I could think of right now is that VictoriaMetrics doenst like the format batch inserts come (VictoriaMetrics/VictoriaMetrics#75 & https://valyala.medium.com/measuring-vertical-scalability-for-time-series-databases-in-google-cloud-92550d78d8ae), but it seems it doesnt even get them...

[isarrider]

_vzlogger_logs.txt
the build is the latest in your repo:
claashk@d51cc3d

[r00t-]

seems to not work with VictoriaMetrics

thanks for testing, but i was not aware that was a software we support, i suggest you re-test with influxdb, and/or create a separate ticket and not spam this MR, thanks.

[claashk]

@isarrider : Did you check out the unified_api_buffering branch?. Master is just tracking the vzlogger master.

[isarrider]

@claashk Yes, I cloned it correctly - but then one has to build the Container from the right directory ;)
It works now - I get no interruption in the database, only sth like:
"InfluxDB response was cannot process influx line protocol data: incorrect influx line "0\nmeasurement,uuid=72f49 [...] 0fd1d5 value=0.000000 1777971560254\n": cannot find Whitespace I in "0"; see https://docs.victoriametrics.com/victoriametrics/in��tegrations/influxdb/"

but maybe thats Victoria-Metrics related... ([...] is not the actual text but just some placeholder)

[r00t-]

@isarrider:

It works now

so the issue you tried to report in your earlier comment was caused by running the wrong code and is resolved?

InfluxDB response was cannot process influx line protocol data: incorrect influx line "0\nmeasurement,uuid=72f49 [...] 0fd1d5 value=0.000000 1777971560254\n": cannot find Whitespace I in "0";

influxdb received a single ascii "0" on a line by itself: 0\n.
that is obviously not valid.
i don't see how the code in InfluxDB.cpp could have created such a line.
can you please increase the loglevel so that this debug print from line 306 is logged:
https://github.com/volkszaehler/vzlogger/pull/704/changes#diff-9e32a66171571cb43247af00f2833c8e231c33e7b5472e9c29c92d50c3c2d5aaR306

		print(log_finest, "request body is %s", channel()->name(), request_body);

and/or re-test with plain influxdb, so we don't need to guess if "victoria-metrics" added that "zero"-line.

[isarrider]

I am busy atm with other things but will let you know when I did this test...