Author | Lightnews

MartinDotNet @martindotnet.bsky.social · 1d

So if you're talking specifically about AWS X-Ray propagation works, that's totally not on otel, or .NET's inbuilt tracing.

Outside of lambda's propagation, it should be fine.

1 1

MartinDotNet @martindotnet.bsky.social · 1d

.NET and Java are the 2 easiest and most robust otel languages, specifically in .NET its baked into the runtime.

Aggregating metrics is interesting, I'm not sure what other tool would have been easier, prometheus would have had the same issue.

1 3

MartinDotNet @martindotnet.bsky.social · 1d

The question would be, have you done it at a scale petabytes per-day, at billions of events per-second.

Context is king. Thats context of the cost to do it (staff, infra), cost to maintain it, opportunity cost of things you're not doing, cost of write over read

It can be done, context is important

1

MartinDotNet @martindotnet.bsky.social · 3d

Not yet... soon(tm)

MartinDotNet @martindotnet.bsky.social · 4d

What you're describing isn’t to do with OpenTelemetry, the monitoring ecosystem maybe.

Its the same as blaming bad application UIs on the NoSQL ecosystem.

1

MartinDotNet @martindotnet.bsky.social · 4d

The end user experience of how the data is queried isnt OpenTelemetry though. OpenTelemetry is the protocol, the datamodel, and the generation of the data. Everything else is backends.

1

MartinDotNet @martindotnet.bsky.social · 5d

I'm not sure why OpenTelemetry is at fault here, Meter's and the underlying instruments you report on are part of what otel provides. How you use them is an issue outside of otel.

1

MartinDotNet @martindotnet.bsky.social · 5d

"Does it work as I expect to" is a query, not necessarily a pre-aggregation (aka a metric).

This is where most people trip up, conflating a thing they want to measure with the mechanism they use to generate the data.

Capturing events and aggregating later provides the same result.

2

MartinDotNet @martindotnet.bsky.social · 5d

I'd go a step further out. The starting point of the investigation is actually "is it all channels that don't work, are they crackly, does the dial turn.

The user experience is the place you start, not monitoring hardware

You don't monitor for CPU spikes, you monitor the customer expectations

1

MartinDotNet @martindotnet.bsky.social · 10d

Thats likely what I'll try now, I loke Traefik as I've used it in docker for a long time. It seems like its just not built for people who don't like containers and k8s.

1

MartinDotNet @martindotnet.bsky.social · 10d

After spending about 3 hours today trying to make it work, I've decided that Traefik without docker/k8s just isn't worth it

Nothing I can do seems to get it to recognize a certificate I've generated, and all the docs are focused around the usecases of acme, or k8s ingress etc

very disappointing :(

1 1

MartinDotNet @martindotnet.bsky.social · 12d

I tried... it just doesnt work that way (you can't rewrite code) it would need something like fody, but I'm not sure that gets access to comments.

Right now, putting a lot of it as attributes in a bindings layer, making it lower touch and out of the main flow.

1

MartinDotNet @martindotnet.bsky.social · 13d

You're absolutely right!

1

MartinDotNet @martindotnet.bsky.social · 18d

"If it takes longer to make your coffee than it does for you build pipeline to complete, you're doing one of them wrong"

- @abangser.bsky.social

4 8 27

MartinDotNet @martindotnet.bsky.social · 26d

This is probably borne out of the idea that POST doesn't include sensitive data in the url or querystring. That stuff being cached in proxies etc.

I'm not saying its right for everything, however as a stance of "POST unless you can justify GET will not leak PII", its not a bad stance.

3

MartinDotNet @martindotnet.bsky.social · 26d

Otel supports both and theres a growing movement to default to http since http2 is becoming the default.

2

Reposted by MartinDotNet

Liz Fong-Jones (方禮真) @lizthegrey.com · Sep 18

For the first time: @honeycomb.io is hiring open roles in Australia!!! We have this senior role open as well as a mid-level role. job-boards.greenhouse.io/honeycomb/jo...

Once we fill these, we will have a thriving APAC team of 5 people: Field CTO, account exec, customer architect, and 2 support.

Senior Technical Support Engineer- AUS

Remote - Australia

job-boards.greenhouse.io

6 32 67

MartinDotNet @martindotnet.bsky.social · Sep 18

I don't know who needs to hear this, but Blazor doesn't use NPM, use that information as you will.

1 10

Reposted by MartinDotNet

Honeycomb @honeycomb.io · Sep 16

It’s only been a couple of days since #o11yDaySFO, and we’re still buzzing.💛 From keynotes to panels to demos, the day showed how AI and observability are shaping the future.

Stay up to date on future events: buff.ly/Qz5wTxH

#honeycomb #observability

2 4

MartinDotNet @martindotnet.bsky.social · Sep 16

That does imply that people do, in fact, have taste. There's mounting evidence to the contrary.

Reposted by MartinDotNet

Honeycomb @honeycomb.io · Sep 9

🚀 The first drop of #AIWeek is here: Honeycomb MCP is GA!

Now with BubbleUp, heatmaps, and histograms, giving humans + AI agents a smarter way to explore and understand systems.

👉 Read all about it here: buff.ly/P5U05Sc

#honeycomb #AI #observability

2 4

MartinDotNet @martindotnet.bsky.social · Sep 7

The big part that's easier is the orchestration of 2 local projects such that they can talk to each other easier, and the connecting of servicebus.

Ultimately, I'm almost trying to get people to not know aspire is there, because its not an Aspire workshop. Ive even removed the servicedefaults.

MartinDotNet @martindotnet.bsky.social · Sep 7

Maybe? This is a telemetry workshop, so its not aspire related really other than we're using the dashboard. The slightly nice bit is that we're using the OpenTelemetry component we've just merged into the toolkit, so forwarding become a little easier.

1

MartinDotNet @martindotnet.bsky.social · Sep 6

Using Aspire for workshop code is a true game changer for the attendees.

Being able to clone the repo and have ports auto-generated and connected means that we can get to the complicated stuff so much faster.

1 8

MartinDotNet @martindotnet.bsky.social · Sep 2

github.com/open-telemet...

This is one of the ones you want to watch.

feat: Add optional Task support for BatchExportProcessor and PeriodicExportingMetricReader by jeromelaban · Pull Request #6379 · open-telemetry/opentelemetry-dotnet

Fixes #2816 Design discussion issue: Partially based on #5778 and #5838 Changes The change adds a useThreads parameter to BatchExportProcessor and PeriodicExportingMetricReader in order to switch b...

github.com