Replay with machine auto-start stopped working

Questions / Help
, ,
1

Hi,

I’m routing requests to specific instances using the fly-replay header. The instances are discovered using the Machines API before routing. Now suddenly (since approx. yesterday), the replay mechanism has trouble waking up the instance it is delegated to.

I use the replay header of the form

Fly-Replay: app=target-app-in-org;instance=specific-instance-name

The logs of the receiving app target-app-in-org then produces many lines of

[info]Starting machine
[error][PM01] machines API returned an error: "machine not found"

followed by

[info]Starting machine
[error][PM01] machines API returned an error: "rate limit exceeded"

after a while. It used to work just fine, so it seems that something has changed. The region is AMS. I wanted to report this before I’m going to destroy and recreate the machines to see, if it fixes the issue.

Maybe combining app and instance arguments isn’t supported any more and I should omit the app name?

3

Status update: destroying and recreating the machines seems to have solved the issue. This is still an unexpected interruption. I wonder if it is a general bug, or if it was maybe related to the networking failure in AMS region last week.

4

I tried some quick tests in ewr using a fairly old Machine there as a replay target and it worked fine (with app= alongside instance=), so it might indeed have been limited to ams…

5

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.