![com docker hyperkit cpu com docker hyperkit cpu](https://pixeljets.com/blog/content/images/2020/10/chart5-1.png)
Googling around previously I found a raft of complaints that Docker was leaking like a sieve under Mohave (10.14) but these were brushed off by the Hyperkit guys - saying it was all OK and just look at the reported real memory usage in the detailed screen. Looked pretty much the same as the first but with a blank Docker stats window. Ten minutes later it crashed (but I cant post a second shot). In addition, although the snapshot shows only 11% CPU, that was up/down to 200% (but consistently around 2% for the combined containers in docker stats). Below is a screen shot when running after about an hour (Extracting ROOT_HAHOG features):Īs you can see Docker itself is reporting (relatively) little mem and CPU usage (as per WebODM diag), the docker real memory usage is apparently relatively low (3.8Gb) but MacOS Activity Monitor main screen is reporting very large usage and high mem pressure. It too also crashed after about an hour (no Docker comms) but I was there (with an active docker/stats window open) and saw some interesting behaviour. I restarted Docker, job remained so I changed to the local node and tried again. I did have access to Docker this time and it showed that only node-adm-1 (which wasn’t used) and broker were running, with worker having exited with “1”, webapp with “128” and db with “139” at the same time. I then tried a 947 photo job on Lightning, but after about 12 hrs it also crashed. After a number of iterations (not good when you are on LTE internet and each pull is 2+GB!) with differing configurations, I assumed not enough memory so bought 2800 Lightning credits and completed that job. Had to force kill Docker, set to factory and reinstall. Couldn’t get any detailed stats as docker commands wouldn’t work. When I tried a “full” paddock (363) docker basically crashed with around 200% CPU. My first couple of tests (~100 20mp photos) were fine. I used the paid installer (thanks - worked well and happy to support) and set Docker to 2/4 CPU and 27/32 GB RAM. While I am not a hard-core “geek” I am a former electrical engineer with pretty broad experience. I have been attempting (mostly unsuccessfully) to use WebODM on my iMac (2016, i5, 32Gb RAM, 10.14.3 Mohave).