Persistent 'Could not Reserve Resource' error in ord

[koz]

My app in ORD is now failing to be deployed because every time I try to run a new deployment I am getting a ‘insufficient resources available to fulfill request’ error. I have tried removing the volume for my app, and creating a new one with hints about the amount of CPU and RAM I need for the app, but even a freshly created volume is failing to schedule.

This is now the third day of me having these errors. If I wait eventually I can get my app to schedule, but since I deploy on push to my github repo I keep getting failures.

Is there some capacity issue in ORD that is causing these heightened failures ?

[koz]

Using the machines API, its making me think there might be some capacity problems, but it looks like pretty much all USA data centers are at negative capacity

curl -s https://api.machines.dev/v1/platform/regions | jq -c '.Regions[] | {"name":.name,"code":.code,"capacity":.capacity}'

{"name":"Amsterdam, Netherlands","code":"ams","capacity":-1174}
{"name":"Stockholm, Sweden","code":"arn","capacity":817}
{"name":"Mumbai, India","code":"bom","capacity":79}
{"name":"Paris, France","code":"cdg","capacity":1709}
{"name":"Dallas, Texas (US)","code":"dfw","capacity":472}
{"name":"Secaucus, NJ (US)","code":"ewr","capacity":-613}
{"name":"Frankfurt, Germany","code":"fra","capacity":-1016}
{"name":"Sao Paulo, Brazil","code":"gru","capacity":1177}
{"name":"Ashburn, Virginia (US)","code":"iad","capacity":-5629}
{"name":"Johannesburg, South Africa","code":"jnb","capacity":251}
{"name":"Los Angeles, California (US)","code":"lax","capacity":-584}
{"name":"London, United Kingdom","code":"lhr","capacity":776}
{"name":"Tokyo, Japan","code":"nrt","capacity":-360}
{"name":"Chicago, Illinois (US)","code":"ord","capacity":-4974}
{"name":"Singapore, Singapore","code":"sin","capacity":-795}
{"name":"San Jose, California (US)","code":"sjc","capacity":-1280}
{"name":"Sydney, Australia","code":"syd","capacity":-170}
{"name":"Toronto, Canada","code":"yyz","capacity":739}

[mayailurus]

Hi… This API has been showing negative values since at least December, unfortunately, and a recent patch to the official flyctl repository described those API outputs as “currently inaccurate”.

(It’s too bad, since it was a great resource while it lasted, …)

[koz]

Well that’s a shame. I did have better luck spinning up machines in DFW and ERW, but its still not reliable that I can deploy a new version and actually be able to schedule my machines

[koz]

Not really a solution, but an update. I now deploy in three regions, and just hope that I can schedule machines in at least one to keep my app functional. So far my faith has been rewarded, but having at least one machine online to process jobs.

This isn’t a great situation to be in, and some clarity about stock-out’s would be really helpful. Even just fixing that endpoint so I had some clarity on availability would be helpful