yangxuanjia

good catch!

@wilsonwang371 try to 1 million kv on one range. 10000 kv is too meek.

@wilsonwang371 I think your test may fall into a misunderstanding. You always prove that your test data looks good in the range under the set preconditions. 1. As an underlying...

@ptabor --max-list-size-bytes=50MB I don't think it's a good settings. Imagine a scenario where the application system has just been launched, and etcd has set --max-list-size-bytes=50MB. At this time, because the...

> @yangxuanjia: Question: Was there much analysis behind the decision to use a 1k batch size? In particular, I'm wondering what happens if we stream each result individually (sort of...

rangeStream we use it for k8s big range to resolve OOM problem. I modify the inner rangeStream version include etcd, grpc-proxy. combine rangeRequest prototype for range and rangeStream. If need,...

@ptabor This bug is difficult to reproduce under unit testing. We reproduced this problem by setting up a cluster environment of physical machines, and then simulating range requests through many...

I make a test for easy test in unit test, to repro the case. ```go package main import ( "bytes" "fmt" "net/http" "sort" "sync" "time" _ "net/http/pprof" go_metrics "github.com/rcrowley/go-metrics" go_bytesize...

use get() to get 300k pods in 100 threads memory OOM:  use getStream() to get 300k pods in 300 thread memory occupy 48G, and last roughly keep stable. ![Screenshot...