nathans wrote:
> Got it. So, next I'm wondering... what is [the module name] doing
> here, in this interface between systemtap/pcp? [...]
systemtap is not a single bundle of instrumentation. It is a tool for
building/running many instrumentation scripts. The unique module name
lets the kernel tell them apart.
> In MMV (and the existing pmdammv, in particular), the basename of
> these files is used to form the first component of the metric
> namespace [...]
That's a most straightforward possibility, though not the most
user-friendly one, as the module names change from run to run.
Perhaps the systemtap script could propose a pmns prefix via an
auxiliary file.
> [...] I guess you'd have to then completely start over again [for
> JSON]
Yes, but the systemtap side of this would be pretty trivial. Printing
JSON strings is easy.
> (for ... reasons?)
Other messages in the thread pointed out some reasons, namely:
- ability to generalize to event traffic, not just sampled metrics
- robustness, by avoiding fragile C code
> [...] I haven't seen any other need for a generic JSON interface
> [...]
In previous notes it was pointed out that JSON is a good encoding for
stats data, because it is exported by other tools like ceph, zabbix,
and many others including
> [...] elasticsearch [...]
So the proposal is to think about a single general JSON PMDA that can
be configured to bridge data from multiple JSON-emitting applications,
including systemtap scripts.
- FChE
|