On Monday afternoon (Dec 4th) we had a severe outbreak of planned maintenance, which resulted in disruption of most MET services between 15:00 and 18:00 UTC. To cut a long story short, someone tried to turn off some old DNS servers which were thought not to be in use, which resulted in DHCP overload and most VMs becoming unresponsive. From what we can see, api.met.no was unavailable for about 1 hour and 39 minutes, although delivery varied from product to product.
The admins have identified the problem, and made some adjustments so that similar incidents should not happen again. Next week they will restart all services in order, after which they will make another attempt to turn off the old DNS servers. We don't expect any downtimes on api.met.no, but if you should experience problems during the following time periods, no action is needed:
In addition they are also changing some network components which are likely to cause some short disruptions in our services. We had about 3 mins downtime on Tuesday (Dec 5th) around 09:30 UTC, although it is difficult to see if it affected the whole API or just our monitoring systems. More maintenance was planned for today, but postponed due to critical weather conditions. The current plan is to upgrade the remaining components on Monday (11th) at 09:00 UTC, when there might be some short interruptions of service. We're sorry for the inconvenience, but hopefully things will be stable during the xmas period.
As for the rest of the API, there have only been a few minor changes the last six months:
Plans for next year include migrating to new servers during February, hopefully with better performance. We are also busy implementing the OGC API EDR standard, which is likely to be the predominant search interface for all MET services in the future and something you might like to read up on.
