FreeBSD jobs occassionally fail when ejecting the disk

[yorickpeterse]

In the last few weeks I've been observing FreeBSD jobs failing when trying to eject a disk, with the error being that the resource is busy. https://github.com/inko-lang/inko/actions/runs/6520144623/job/17707484871 is one such job where this happened:

2023-10-14T22:02:10.0083480Z Waiting for partitions to activate
2023-10-14T22:02:10.0365870Z Formatting disk2s1 as MS-DOS (FAT32) with name RES
2023-10-14T22:02:20.8279870Z 512 bytes per physical sector
2023-10-14T22:02:20.8280650Z /dev/rdisk2s1: 76594 sectors in 76594 FAT32 clusters (512 bytes/cluster)
2023-10-14T22:02:20.8282080Z bps=512 spc=1 res=32 nft=2 mid=0xf8 spt=32 hds=16 hid=2048 drv=0x80 bsec=77824 bspf=599 rdcl=2 infs=1 bkbs=6
2023-10-14T22:02:20.9370120Z Mounting disk
2023-10-14T22:02:24.9493860Z Finished partitioning on disk2
2023-10-14T22:02:25.0276190Z [command]/usr/bin/sudo umount /Volumes/RES
2023-10-14T22:02:26.1238370Z [command]/usr/bin/hdiutil detach /dev/disk2
2023-10-14T22:02:27.0086020Z hdiutil: couldn't eject "disk2" - Resource busy
2023-10-14T22:02:27.0565870Z 
2023-10-14T22:02:27.0569630Z /Users/runner/work/_actions/cross-platform-actions/action/v0.19.1/webpack:/cross-platform-action/node_modules/@actions/exec/lib/toolrunner.js:574
2023-10-14T22:02:27.0571900Z                 error = new Error(`The process '${this.toolPath}' failed with exit code ${this.processExitCode}`);
2023-10-14T22:02:27.0573230Z ^
2023-10-14T22:02:27.0573860Z Error: The process '/usr/bin/hdiutil' failed with exit code 16
2023-10-14T22:02:27.0576500Z     at ExecState._setResult (/Users/runner/work/_actions/cross-platform-actions/action/v0.19.1/webpack:/cross-platform-action/node_modules/@actions/exec/lib/toolrunner.js:574:1)
2023-10-14T22:02:27.0579510Z     at ExecState.CheckComplete (/Users/runner/work/_actions/cross-platform-actions/action/v0.19.1/webpack:/cross-platform-action/node_modules/@actions/exec/lib/toolrunner.js:557:1)
2023-10-14T22:02:27.0584190Z     at ChildProcess.<anonymous> (/Users/runner/work/_actions/cross-platform-actions/action/v0.19.1/webpack:/cross-platform-action/node_modules/@actions/exec/lib/toolrunner.js:451:1)
2023-10-14T22:02:27.0586440Z     at ChildProcess.emit (node:events:513:28)
2023-10-14T22:02:27.0587160Z     at maybeClose (node:internal/child_process:1100:16)
2023-10-14T22:02:27.0588120Z     at Socket.<anonymous> (node:internal/child_process:458:11)
2023-10-14T22:02:27.0588840Z     at Socket.emit (node:events:513:28)
2023-10-14T22:02:27.0589440Z     at Pipe.<anonymous> (node:net:301:12)

A restart of the job typically fixes it, though sometimes a second or third retry may be necessary. I'm not sure yet how to reliably reproduce it, as it happens sporadically.

[jacob-carlborg]

I've started to see this myself as well. Also for OpenBSD. Not sure what to do about it. Catch the error and try the command again perhaps.

[yorickpeterse]

The last week or so I've been noticing more and more of these failures. Sometimes a single retry is enough, other times I have to retry several times before the job succeeds. Some sort of retry mechanism with a small wait time (e.g. 5 times with a 30 sec wait time) would be much appreciated 😃

[jacob-carlborg]

Looks like it was possible to fix/mitigate this issue by retrying detaching the disk: https://github.com/cross-platform-actions/action/actions/runs/6749533591/job/18350090001#step:3:236. Fixed in https://github.com/cross-platform-actions/action/releases/tag/v0.21.1.

[yorickpeterse]

@jacob-carlborg Thanks! I'll give it a try! 🎉