山岚
An internal test shows that, with 100ms latency, some of the ranges may cost up to 40 secs to be uploaded. This may cause the importer backend failed because it...
An experiment shows that, with RTT growing, the throughput of raw TCP would be limited. when RTT doubles, throughput would become half. Detailed info: ``` ./test-result/ping-0ms.log : [ 4] 0.00-10.00...
So I guess the final artifact should be a SQL stmt like `SHOW BACKUP INFO IN $storage`?
Maybe https://github.com/tikv/pd/issues/4797
Goroutine dumping: [stuck-goroutines.txt](https://github.com/pingcap/br/files/6227313/stuck-goroutines.txt)
Seems one call to backupClient.Recv stuck. Goroutines waiting chain: 1. for files := range filesCh (client.go:454) (the key!) 2. for err := range errCh (client.go:485) (the key!) 3. eg.Wait() (cilent.go:477)...
TiKV log: [br-stuck-tikv-log.tar.gz](https://github.com/pingcap/br/files/6262899/br-stuck-tikv-log.tar.gz)
I'm afraid this was caused by random pause of virtual machines / containers...(Or, GC?) Anyway, let's try to reproduce this.
For making sure there isn't a deadlock or something, I set max unblock time to 2s and run it about 4,000 times, but no `WaitGroup did not unblock after maximum...
This time `WaitGroup unblocked before minimum duration`...