William Chan

I haven't thought too much about how to achieve the suggested goals I laid out to @domenic that are important for high performance networking. Let me sketch out some ideas...

> > Have a lower layer ReadableByteStream and WritableByteStream classes. These need to support byte stream and piping APIs that mirror POSIX read()/write()/splice(). They do not contain any affordances for...

 Yeah, there's fundamentally a sucky choice to be made here. I can see both sides of the debate. I'm not as hopeful that you can layer on...

Sorry, I forgot to explain how it breaks the compositional abstraction. Indeed, write() would do that for you, but there's a performance consequence in how you invoke write() that is...

Yep, you've nailed the tradeoffs here. The stream implementation can try to be smart and use heuristics to trade off between efficiency and latency. Note that the specific byte sizes...

Our metric is called `k8s_http_requests`, leading to the following PromQL: `logging_googleapis_com:user_k8s_http_requests{request_status=~"5\\d\\d"}`. Note the escaping of the backslash, since the `\d` is the regex escape sequence.

Well, we'd like to alert on 5xx error codes other than just 500, like 502 Bad Gateway or 503 Service Unavailable. I can create a separate alert for each one,...

Yeah, I was wondering if that would get fixed with official support. But if my diagnosis is correct, the root problem is that the different collector Pods in the DaemonSet...

Indeed, GKE Autopilot sets the limit to the request, which is why we get OOMKilled. That said, I believe the resource & limit configuration is a bit of a red...

Hm, I may have misunderstood when the docs says that VerticalPodAutoscaler is enabled by default. On a re-read, I think maybe it just means that's enabled as a cluster feature,...