Alexander Prinzhorn

Can you somehow verify this by switching between two mitmproxy instances? I assume, if our theory is correct, you'll see the requests in the "old" mitmproxy that isn't actually used...

I'm not sure why it's that way, but it happens here https://github.com/mitmproxy/mitmproxy/blob/8178f6c77b715cd342f8be161d5ca5da8938645c/mitmproxy/addons/dumper.py#L110-L113 but all calls to `_echo_message` are already guarded with `if ctx.options.flow_detail >= 3:` so the `else` will _never_...

I changed my mind, sort of. `flow_detail = 4` is simply undocumented :thinking: . You can use it ~as a workaround~ :+1:

I've made a PR for the docs and updated this to be a feature request to make the `70` configurable when using flow_detail 3

Just in case Shopify changes something on their end and it won't be reproducible anymore, here's verbose logs with `--set proxy_debug -vvv` [shopify.log](https://github.com/mitmproxy/mitmproxy/files/7877694/shopify.log)

Thanks for all the great PRs! > ideally we should also handle the case if a mode change after startup fails. That's a tricky problem and one that I brought...

> Listener State Did you see my comment on the original PR? https://github.com/mitmproxy/mitmproxy/pull/5393#issuecomment-1153666100 I know this is yet another discussion, but I just wanted to bring that up one last...

> If we don't need "starting" and "stopping" as status, we could also just trigger `ServersUpdated` once after `setup_servers` ran. From a UX point I'd love to have these states....

> So we need/want pull anyways. Does that make sense now? I still can't really follow :smile: . What keeps mitmweb from opening a WebSocket which would give you the...

Alright, I think I see the problem here. I just looked at the dev tools in mitmweb. I had no idea about it's architecture. It looks like mitmweb fetches the...