Phoenix Plug.Cowboy adapter support #133
Conversation
bryannaegele
added
scope-ci
opentelemetry_phoenix
elixir
and removed
opentelemetry_phoenix
elixir
labels
| ## Usage | ||
|
|
||
| In your application start: | ||
|
|
||
| def start(_type, _args) do | ||
| :opentelemetry_cowboy.setup() | ||
| OpentelemetryPhoenix.setup(adapter: :cowboy2) |
There was a problem hiding this comment.
Is there a way to detect, or just assume, this? Aren't the vast majority of uses cowboy2, why not make it that it is cowboy2 unless you supply an adapter option?
There was a problem hiding this comment.
We'd have to get a config added to specify the endpoint and then parse through the config to find which adapter is in use for it. We can't assume when there are multiple endpoints running because we're setting up the handlers based on the telemetry prefix. The other option is we have to parse through the conn on every request to try to find it which seems a little a
No matter what, the user is going to have to install server instrumentation for the adapter they're using. I don't have qualms with making the user specify which one to use. It's also likely Bandit will become the goto within a year or two. Its performance is substantially better and has support in Phoenix 1.7. If it holds up, I could see it becoming the default in a future release.
We could make it the default, but if they don't add and setup cowboy instrumentation then it just won't work at all. Felt like a breaking change if we went that route. Or this lib needs to package up multiple adapters and install them whether they use them or not. I mean, Ecto requires you to say which adapter you're using through config. With Ecto repos, you're adding specific repos to the supervision tree to start them. Phoenix does the same with JSON library tied to an endpoint, but the endpoint again is added to the supervision tree. So at some place there's config happening and they're added to the start or supervision tree.
I dunno.
Co-authored-by: Tristan Sloughter <t@crashfast.com>
* fix: redix missing dep * chore: add sc dep * chore: use semantic conventions in phoenix
Add instrumentation for Nebulex, a distributed cache library. This library provides solid telemetry support for this initial implementation. Caching implementation is mostly based on in-memory storage (like ETS) and RPC calls for distribution (via OTP libraries, like :erpc). AFAICT, there is not much specifics for how to translate into Semantic Attributes: those caches are not quite a DB, except maybe for the one which implements the storage; the RPC can't be reliably captured either. Given the above constraints, this initial implementation instruments the library via custom attributes (namespaced as `nebulex.*`). It's not 100% clear the behaviour of OTel for actual distributed caches - from my tests, that may create some orphan spans. I think that's fine as first release. Nebulex follow the patterns of Ecto, so this instrumentation follows a similar pattern of OpentelemetryEcto. It does include a `setup_all/1` function for convenience, that leverages the :init events Nebulex emit on process start. Co-authored-by: Tristan Sloughter <t@crashfast.com>
|
Really bad rebase merge so reopening with #144 |
Adding optional adapter support for
Plug.Cowboyto continue the existing cowboy span rather than starting a new one.Bandit.PhoenixAdaptersupport will require a telemetry instrumentation lib for it.The phoenix example app is also updated to be more friendly for user experimentation and iteration by removing the docker app build as the default way to run it.