I've noticed that the Regex engine for timelion labels does not quite work as expected. For example I've tried plugging the regex in an online visualizer such as Regex Tester, it will match what I expect and then I'll plug it into timelion and get something absolutely different.
For example on https://www.regexpal.com/ with this input:
q:kubernetes.pod.name:qvp6rolm* AND _exists_:kubernetes.container.memory.usage.bytes > kubernetes.pod.name:qvp6rolm-xis88m7q-mycontainer-0 > avg(kubernetes.container.memory.usage.bytes)
and this regex:
I expect the matched group to be:
However, on timelion the output is:
q:kubernetes.pod.name:qvp6rolm* AND _exists_:kubernetes.container.memory.usage.bytes > qvp6rolm-xis88m7q-quorum-node-0 avg(kubernetes.container.memory.usage.bytes)
To fix the regex for timelion I have to change the regex to:
Is there a utility that I can use to rapidly prototype these regexes in order to visually validate them before using them in timelion has a high overhead when recomputing the regex?
Also, could we perhaps think about giving access to fields in the time series and just using a templated string that references time series fields instead of a regex? For example if I split on name, it would be convenient to just label each series with the name field.