Feature request - expose environmental variables to configuration file about carbon-c-relay HOT 8 CLOSED

Just noticed your commit for capture groups (not production ready). The memory consideration will probably not be an issue on the clients in our environment - the flow of metrics will be a constant.

I shall give it a go and report back.

The variables thing would still be nice though.

from carbon-c-relay.

re: patches, beware that I'm very picky about code ;) For the statistics interval change it's a matter of passing on a (command line flag fed) value down to the main loop. It does second sleeps, so, it should be a matter of replacing the hardcoded 60 with the value passed on. Not a big change, feel free to open a separate feature request for it.

from carbon-c-relay.

Is it really necessary to read all environment environment variables? Hostname is already known to the relay (and can be overridden using -H) so that would be slightly easier to use.

from carbon-c-relay.

no, I don't think that all environment variables are needed.How do I use
the variable Hostname set through -H?

I was trying to avoid a situation like:

Server maxwell

aggregate
    *.system.cpu-[0-9]+.cpu-idle
  every 10 seconds
  expire after 5 seconds
  compute average write to
    maxwell.vbox.cpu-all.average

Server edison

aggregate
    *.system.cpu-[0-9]+.cpu-idle
  every 10 seconds
  expire after 5 seconds
  compute average write to
    edison.vbox.cpu-all.average

Server beyonce

aggregate
    *.system.cpu-[0-9]+.cpu-idle
  every 10 seconds
  expire after 5 seconds
  compute average write to
    beyonce.vbox.cpu-all.average

and so on.

With the back references I don't need that and I can just do:

aggregate
     (.*).system.cpu.[0-9]+.cpu.([a-z]+)
   every 10 seconds
   expire after 5 seconds
   compute average write to
     \1.system.cpu.avg.\2
   ;

But I am wary that as you said, this is not well tested yet.I will
provide feedback after it has gone through our load test environment.

I also have the unhappy situation where I get a stream of metrics on
each client like:

127.0.0.1.memcached.metrics_cache.df.cache.used

Which is annoying as I want:

maxwell.vbox.memcached.metrics_cache.df.cache.used

Having the hostname (fqdn) available would save a considerable amount of
Puppet cleverness.

Regards and thanks, Carbon C Relay has made my day job easier.

from carbon-c-relay.

Sorry, gentlemen, for interrupting your conversation, but BTW - you can use built-in collectd aggregator plugin for aggregating CPU stats. Something like

LoadPlugin aggregation

<Plugin "aggregation">
  <Aggregation>
    Plugin "cpu"
    Type "cpu"

    GroupBy "Host"
    GroupBy "TypeInstance"

    CalculateSum true
    CalculateAverage true
  </Aggregation>
</Plugin>

works like a charm...
Just my 5 cents.

from carbon-c-relay.

deniszh
5c well spent.
I did not notice the collectd aggregation plugin.
Ta.

from carbon-c-relay.

Would it be worth to implement something like \H in the replacement code, or can we sufficiently get away with group matches and different input here?

from carbon-c-relay.

assuming all is fine here, please reopen if I overlooked something

from carbon-c-relay.