[BUG] metrics with utf8 strings in path are not displayed

[mcikosos]

Describe the bug
metrics with utf8 strings in path are not displayed in graphs

To Reproduce
just send any metric with national chars in it

ex.
echo "local.random.gówno 5 date +%s" | nc ${SERVER} ${PORT}

Expected behavior
metrics show in graphs

Screenshots
no point

Environment (please complete the following information):

• OS flavor: Ubuntu
• Graphite-web version: latest docker - graphiteapp/graphite-statsd
• Django/Python version: latest docker - graphiteapp/graphite-statsd
• Setup type latest docker - graphiteapp/graphite-statsd

[mcikosos]

When I change metric path in graphite web - using edit to change it to 'local.random.g?wno' - the data are shown

[deniszh]

Hi @mcikosos ,

IIRC UTF-8 characters can work in tags names/values (because tags are hashed internally), but UTF-8 support in metrics itself is depending on your OS / filesystem setup. I mean, I'm not saying that's not a bug - and if someone can solve it in generic and backward compatible way - I would be very glad.

[piotr1212]

I found a few issues with unicode chars in Graphite-web, carbon on the other hand does work.

This fixes two piotr1212@17e23ef but I never got around of making a PR. I think there was also one in the carbonlink code. And last, the images rendering code cannot handle unicode (metricnames in the legenda). The last one is the hardest to fix but probably the least important one. There is probably some more, potential places to look are str() calls

[mcikosos]

IIRC UTF-8 characters can work in tags names/values (because tags are hashed internally), but UTF-8 support in metrics itself is depending on your OS / filesystem setup. I mean, I'm not saying that's not a bug - and if someone can solve it in generic and backward compatible way - I would be very glad.

Well I thought it is a BUG because carbon/whisper etc. works ok but if it was not supported from the start I can change it to "feature" ;)

[mcikosos]

@piotr1212 can You make a PR? So some day it will be merged to master.

[piotr1212]

Sorry, I don't expect to anytime soon.

[deniszh]

Oh, cool, I didn't know that, @piotr1212 !
PR is welcome ofc, I will pick this up if nobody else.

[deniszh]

@mcikosos : Fix is merged in master. UTF8_METRICS should be enabled in config (disabled by default because of possible negative performance consequences).

[mcikosos]

I will check this ASAP