austinlparker

@austinlparker@hachyderm.io

director of open source @honeycomb.io

opentelemetry maintainer

devrel, author, etc. eccentric.

This profile is from a federated server and may be incomplete. Browse more on the original instance.

mcc, to random
@mcc@mastodon.social avatar

Looking into how to set up a bluesky server.

Noticed this bit here, in the documentation. That's an interesting way to spell "cannot"

austinlparker,

@mcc you might find this GitHub issue interesting in terms of what the PDS does — iirc, there’s at least one independent fork written in go floating around, but I don’t think it’s up to date, but it might be interesting? https://github.com/bluesky-social/atproto/discussions/2350

also the main pds source is in indigo, not in the pds repo - https://github.com/bluesky-social/indigo/tree/main/pds

austinlparker,

@mcc I believe the thing what’s in docker is a typescript PDS implementation which is based off https://github.com/bluesky-social/atproto

one thing you can do is throw in some OpenTelemetry environment variables in and run a local collector and it should start spitting out traces showing what’s going on

austinlparker,

@mcc that said the whole thing is still pretty janky in terms of “ease of use for self-hosting”, especially if you don’t want to use docker

paigerduty, to random
@paigerduty@hachyderm.io avatar

only 50 pages into Learning OpenTelemetry by @austinlparker & @tedsuo but have already highlighted so many choice quotes like

"To be more blunt, your ability to understand a software system is ultimately a cost-optimization exercise." 💯 🤑 💻

also came across a new-to-me word "presages"

austinlparker,

@paigerduty @tedsuo presages is one of my favorite words

austinlparker,

@paigerduty probably

hazelweakly, to random
@hazelweakly@hachyderm.io avatar

Pretty much all of the woes of distributed tracing are caused from the mismatch of the mental model of distributed tracing that makes sense vs the one that can be built easily:

The model that makes sense is "lazily built and incrementally fleshed out call graph with late-binding updates of attributes as discovered"

But the way that makes sense to implement it is "strict call-stack semantics with fire-and-forget frozen rows of data into an append only data store"

austinlparker,

@hazelweakly ah, segments vs spans, aka “who do you want to pay for this”

if you want to pay at query time, segments (and then you wind up with stuff that looks more like metrics in terms of cardinality)

if you want to pay upfront, spans.

austinlparker, to random

Hey, it’s here! In print! Learning OpenTelemetry now available wherever fine books are sold. Find more information at https://learningopentelemetry.com

austinlparker, to random

I did the thing

https://aparker.io/2024/03/dont-work-for-projects-that-dont-have-open-governance/
https://hachyderm.io/@austinlparker/112144651656988834

austinlparker, to random

Want to chat about OpenTelemetry? I’ll be at All Things Open this week, so look me up!

austinlparker, to random

I heard that Splunk was expensive, but this is ridiculous!

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • megavids
  • khanakhh
  • mdbf
  • ethstaker
  • magazineikmin
  • GTA5RPClips
  • rosin
  • thenastyranch
  • Youngstown
  • InstantRegret
  • slotface
  • osvaldo12
  • kavyap
  • DreamBathrooms
  • JUstTest
  • Durango
  • everett
  • cisconetworking
  • normalnudes
  • tester
  • ngwrru68w68
  • cubers
  • modclub
  • tacticalgear
  • provamag3
  • Leos
  • anitta
  • lostlight
  • All magazines
    •