October-16th-2018, 06:54 PM
Hi Steve,
pulling the image on a Raspberry Pi 3B CPU (on netPI then also) takes longer than on other CPUs. This is because of the bottleneck (serial interface) between the CPU and the SD card used as memory storage and the small cache the CPU owns. Also it depends on the traffic that is ongoing on Docker hub, if you pull an image from there. Then also it has influences if you have Containers actively running already and of course it depends on how big your image is. All those parameters influence the Container load time.
Additionally I have a remark that since firmware 1.1.2.0 we improved the load time by removing a permanent logging of internal Docker messages. Please verify that you have this firmware or higher actively running on your netPI.
If you want to get rid of depencencies from Docker Hub, I recommend to set up a local registry anyway. Then at least this timing is constant.
And if I would summarize my experience till today I would say that seeing this timeout is rare and that starting a container with implicit pull works in 95% of all cases. The timeout is not caused by Docker running on netPI by the way, but due to a Web-Browser <-> portainer.io timeout.
pulling the image on a Raspberry Pi 3B CPU (on netPI then also) takes longer than on other CPUs. This is because of the bottleneck (serial interface) between the CPU and the SD card used as memory storage and the small cache the CPU owns. Also it depends on the traffic that is ongoing on Docker hub, if you pull an image from there. Then also it has influences if you have Containers actively running already and of course it depends on how big your image is. All those parameters influence the Container load time.
Additionally I have a remark that since firmware 1.1.2.0 we improved the load time by removing a permanent logging of internal Docker messages. Please verify that you have this firmware or higher actively running on your netPI.
If you want to get rid of depencencies from Docker Hub, I recommend to set up a local registry anyway. Then at least this timing is constant.
And if I would summarize my experience till today I would say that seeing this timeout is rare and that starting a container with implicit pull works in 95% of all cases. The timeout is not caused by Docker running on netPI by the way, but due to a Web-Browser <-> portainer.io timeout.
„You never fail until you stop trying.“, Albert Einstein (1879 - 1955)
