bwplotka.dev
Senior SWE TL Google | ex Principal RedHat | Prometheus, Thanos OSS maintainer | CNCF TAG Observability TL | Efficient Go 📖 OReillyMedia
70 posts
723 followers
39 following
Prolific Poster
Active Commenter
comment in response to
post
Well anyone can do it easily with go bench and benchstat these days, exported to @prometheus.io or sheet graphing.
Maybe you meant benchmark.vitess.io ?
comment in response to
post
Now you got my attention
comment in response to
post
Context: github.com/prometheus/p...
comment in response to
post
Tdd, no? :p
Thumb driven development
Virtual democracy
comment in response to
post
Thanks! Yea this is really generous from Go side, hope users realise that (:
comment in response to
post
Wow what an investigation! Sounds like some library was just using runtime internals unsafely 🤔
Curious why that's a Go release blocker (I didn't dive very deep into discussion)
comment in response to
post
HN: news.ycombinator.com/item?id=4277...
comment in response to
post
Agree, I never know what to say here, choose randomly (:
comment in response to
post
Yup. The complex type handling feels like the most interesting and relevant now, motivator against dots in unquoted metric names.
Perhaps actually trying to propose something there with dots (or check alternative chars) would be beneficial to progress here and reduce confusion (e.g. on why)
comment in response to
post
The challenge is to explain why you need those for dots 🙈
comment in response to
post
Ack, hope its not like you just don't want to admit it to be nice! 😄
I would be ok with that it would be good for me to practice "loving the no" open.spotify.com/episode/4VLh...
comment in response to
post
That's related to my PR? (:
comment in response to
post
Ah nice! Yea, we could make this a pattern. However it feels it would need to handle load balancing too (supporting multiple Prometheus-es), which touches "dynamic scrape target distribution problem" that is generally a bad idea (less reliable)
Somewhat relevant to opentelemetry.io/docs/kuberne...
comment in response to
post
...there's even a OSS metric collection operator we maintain at Google with that optimization 🙃
github.com/GoogleCloudP...
Another solution is to ofc have a cluster service discovery that pushes the targets to node agents, but pure (optimized) daemonset usually scales enough!
comment in response to
post
But what is it? It's not a function name it's some garbage annotation from generated code 🤔
comment in response to
post
However, not going to lie... the remote dev-env company prioritizing running your env locally is kind of funny? 🙃 Full circle.
comment in response to
post
XD
comment in response to
post
This is annoying consequence of the current "verification" model. I moved to my domain recently so I created another account on the old name just in case.. ):
comment in response to
post
I think there are both scalability and social reasons for keeping low char limits, but it's sometimes challenge yes!
comment in response to
post
There's no char limit on mastodon, or just longer?