Avi Kivity

@denesb looks like a reader problem

Is there a test for this?

What if a node in some region (or the entire region) is down? We may need to add a new node in our region, but now we can't, and the...

Now it's clear. schema_tables.cc is littered with calls to coroutine::parallel_for_each(), and with a large number of inputs this can be expensive.

I guess we can litter them with maybe_yield() as the first statement. The alternative is to add yields in coroutine::parallel_for_each(), though that penalizes the common use case of limited concurrency....

Looks good. Is there a dtest for this?

@cmm will you update the branch?

I'll regenerate the image. I'm not happy about packaging a release candidate though.

Is node_exporter actually vulnerable? It says something about PEM data, which is TLS related. Is it doing anything in that area?

I think we poll without TLS anyway. Let's wait for 1.4.0 to be released.