since 7 am. Slow down of the server rendered requests for downloading a new application stack
into timeouts. As a result more and more Meteobridges were caught in a loop asking for the
application stack, which did make things worse.
Root cause are problems of my provider, who surprised me with a significant degradation
of VM performance. Why that is still under investigation.
To get things flying again I did a rework how application stack is individualized (fingerprinted)
for each Meteobridge and reduced number of concurrent HTTP sessions. This needed some
coding and testing and it took until now to have it stable again.
I hope the VM will be soon back to its old performance. The good thing is that the new
code is even more efficient and it now can feed many more Meteobridges
Sorry again, for the hassle. If your Meteobridge is showing errors about not matching hardware
or fingerprints, please reboot. Your Meteobridge has grabbed an early application stack of
today that was not built the right way.