IPNI verification timeout may be too low for APAC storage providers

[TippyFlitsUK]

Problem

APAC-based SPs are seeing intermittent failure.timedout on data storage checks while their nodes are healthy and retrievals pass fine. We've observed IPNI verification times exceeding the 60s limit for providers in this region, resulting in steady ~9% failure rates.

Context

We have 4 APAC SPs trying to get on the approved list. With ~500ms round-trip between Europe and APAC at each stage of the IPNI verification flow, the 60s window is marginal for this region. European SPs with 10-50ms latency are not affected.

Suggestion

Raise the IPNI verification timeout (e.g., 120s) or make it configurable, so that healthy APAC SPs aren't systematically disadvantaged by geography.

[BigLep]

I'm surprised that 500ms geographical latency makes that big of a deal over the course of 60 as I didn't think there were that many requests between the APAC SP and IPNI:

• SP Announces advertisements
• IPNI HTTP GETs advertisement chain from SP
  (unless there are many HTTP GET calls? )

[TippyFlitsUK]

Our APAC friends are having to jump through quite a few hoops to get stable connectivity with the GFC right now, which adds a fair bit of setup complexity.

Hopefully @beck-8’s recent work with non-standard ports helps smooth this out and reduce the amount of workaround needed 🤞

Given the connectivity issues we’re seeing (especially in APAC), I think a 90s timeout is a good compromise 👍
It should give enough headroom for higher latency / network variability without letting things hang unnecessarily.