aspire logs with no arguments is really slow

[davidfowl]

[davidfowl]

I reproduced this locally in /Users/davidfowler/dev/git/dogfood/mytsapp against a running AppHost and measured both commands 3 times.

aspire logs (no args):

• run1: 5.292s
• run2: 5.286s
• run3: 5.294s
• avg: 5.291s

aspire logs --follow:

• measured time-to-first-log-line (process started, then stopped right after first output)
• run1: 60.7ms
• run2: 60.3ms
• run3: 61.3ms
• avg: 61ms

So the no-args path consistently takes ~5.3s before completion, while --follow starts streaming almost immediately (~60ms).

[JamesNK]

Now logs are fetched in parallel. TestShop takes 5 seconds to display logs instead of 30+ seconds.

I think the delay at this point is coming from getting logs from DCP.

How about:

1. Add a "getting logs" status in console UI so you see it doing something rather than a blank delay.
2. DCP folks look at perf of getting logs.

@davidfowl @karolz-ms @danegsta

[karolz-ms]

I will take a peek

[karolz-ms]

This is a bug in DCP all right. Applies to containers only. When capturing container logs in non-follow mode, instead of reading all the logs that we have and immediately returing, we keep waiting for more logs to arrive, with a short timeout (5 seconds). This is complicated a bit by the fact that capturing container logs is somewhat expensive and we do it on-demand, asynchronously, and in a way that shares the burden between requests.

I am working on a fix.

[karolz-ms]

Fixed in 13.2 branch, backmerge to main should occur shortly