Telegraf looks like a very nice way to collect metrics for monitoring purpose, with quite a lots of way to extend it.
However, my experience with real-life supervision is that sometimes there are checks which can be costly, hence which are not scheduled too often (say every 5 minutes), for which operators who solved the issue would like to “force recheck” to get the alert back to green.
One telegraph process can collect many type of metrics using plug-ins, so “forcing check” by restarting telegraph process would work, but it would be quite inelegant, not to mention that it would re-trigger all check,s which can be costly.
I’m looking for a solution to send to Telegraph process (maybe thru its listening socket) a message to ask it to collect data from a given plugin.
Is it possible to do it (I haven’t found any clue of it in current documentation) ?
Thanks for your quick answer, I just created the following feature request:
PS: sending a SIGHUP to force telegraf to reload its configuration seems to me to be relatively the same as restarting telegraf: it’s going to force gathering of all metrics, isn’t it? I need a way to force gathering of only a few metrics, for performance purposes.
