orbstack
Server Connections Interrupted
Describe the bug
The last few days I've been working for 8+ hours straight, using OrbStack throughout the day. Occasionally, and seemingly without a pattern, I will see the OrbStack mounts on MacOS fail. When this happens, my OrbStack machines + docker containers are unusable. After 30-120 seconds, this issues seems to resolve itself, which is interesting, but the unpredictability of the disconnect is problematic.
To Reproduce
- Run OrbStack
- Just use the MacBook as per usual and within an hour I almost always see this disconnect
Expected behavior
Ideally this disconnect should never occur, thus never interrupting the OrbStack resource connectivity.
Diagnostic report (REQUIRED)
OrbStack info: Version: 1.6.4 Commit: 90e832a0f4c6787bd028dc4efcd9c98cd5a2bc3d (v1.6.4)
System info: macOS: 14.5 (23F79) CPU: arm64, 14 cores CPU model: Apple M3 Max Model: Mac15,11 Memory: 36 GiB
Full report: https://orbstack.dev/_admin/diag/orbstack-diagreport_2024-07-17T14-00-12.003741Z.zip
Screenshots and additional context (optional)
No response
Can you try https://cdn-updates.orbstack.dev/exp/OrbStack_v1.6.4-14-gba24e31ab_17206_arm64.dmg and send a new diagnostic report when this happens?
Can you try https://cdn-updates.orbstack.dev/exp/OrbStack_v1.6.4-14-gba24e31ab_17206_arm64.dmg and send a new diagnostic report when this happens?
Yes I can - updated just now, I'll watch it throughout the day today
Any results on the new build?
@kdrag0n so I updated to the new build on July 18 (about 6 days ago) and since the update, I have not seen the issue anymore. It looks like the change you provided seemed to add some major stability.
Odd, there should be no relevant changes other than additional debugging info. Can you revert to stable v1.6.4 to confirm?
happened to me just now, after which orbstack stopped functioning at all. The gui is stuck in a loading state, the CLI panics, containers offline.
panic: timed out waiting for services to start [recovered]
panic: timed out waiting for services to start [recovered]
panic: timed out waiting for services to start
goroutine 1 [running]:
github.com/orbstack/macvirt/scon/cmd/scli/cmd.RecoverCLI()
github.com/orbstack/macvirt/scon/cmd/scli/cmd/utils.go:43 +0x78
panic({0x1011849c0?, 0x1400020c040?})
runtime/panic.go:785 +0xf8
github.com/orbstack/macvirt/scon/cmd/scli/scli.init.OnceValue[...].func5.1()
sync/oncefunc.go:54 +0x78
panic({0x1011849c0?, 0x1400020c040?})
runtime/panic.go:785 +0xf8
github.com/orbstack/macvirt/scon/cmd/scli/scli.check(...)
github.com/orbstack/macvirt/scon/cmd/scli/scli/scli.go:12
github.com/orbstack/macvirt/scon/cmd/scli/scli.init.func2()
github.com/orbstack/macvirt/scon/cmd/scli/scli/scli.go:19 +0x88
github.com/orbstack/macvirt/scon/cmd/scli/scli.init.OnceValue[...].func5()
sync/oncefunc.go:57 +0x74
sync.(*Once).doSlow(0x1400017db88?, 0x100f17168?)
sync/once.go:76 +0xc8
sync.(*Once).Do(0x0?, 0x0?)
sync/once.go:67 +0x24
github.com/orbstack/macvirt/scon/cmd/scli/scli.init.OnceValue[...].func6()
sync/oncefunc.go:62 +0x3c
github.com/orbstack/macvirt/scon/cmd/scli/cmd.init.func32(0x10157f8a0, {0x1015fd580?, 0x4?, 0x100ff436b?})
github.com/orbstack/macvirt/scon/cmd/scli/cmd/run.go:156 +0x7c
github.com/spf13/cobra.(*Command).execute(0x10157f8a0, {0x1015fd580, 0x0, 0x0})
github.com/spf13/[email protected]/command.go:916 +0x648
github.com/spf13/cobra.(*Command).ExecuteC(0x10157a540)
github.com/spf13/[email protected]/command.go:1044 +0x320
github.com/spf13/cobra.(*Command).Execute(...)
github.com/spf13/[email protected]/command.go:968
github.com/orbstack/macvirt/scon/cmd/scli/cmd.Execute(...)
github.com/orbstack/macvirt/scon/cmd/scli/cmd/root.go:36
main.runCtl(0x50?)
github.com/orbstack/macvirt/scon/cmd/scli/main.go:143 +0x1dc
main.main()
github.com/orbstack/macvirt/scon/cmd/scli/main.go:44 +0x84
I've also been seeing this for months through various OrbStack and MacOS updates, but it is very pronounced currently and happens a couple of times a day. Sometimes OrbStack can't recover after restart it ("Service Not Running") and a whole machine restart is needed to get it running containers again.
If I ignore the "Server Connections Interrupted" popup and try to send a bug report it fails after a minute or so with "The data couldn't be read because it isn't in the correct format."
M1 MacBook Pro MacOS 15.3.1 OrbStack 1.9.5 (18849)
I'm getting this too. Seems to have increased significantly, recently, and I can't figure out why. It does seem especially more likely when I'm trying to import a database or otherwise handle larger jobs/processes. I tried to "Submit Bug Report", but I get the same thing as @bmtcril: "The data couldn't be read because it isn't in the correct format."
That said, I did notice this last time in Activity Monitor that my computer's memory was being almost completely used by a Google Drive process. I know there was a somewhat-recent change to Orbstack on Mac's where the memory usage is supposed to be "right-sized" or dynamically using only what it needs. My suspicion–based on just those two small clues– is that this has lead to Orbstack sometimes getting caught in a situation where it can't appropriate enough memory.
Does that sound plausible? After I quit Google Drive and allowed the memory usage to come back down, I restarted Orbstack and everything was peachy.
"The data couldn't be read because it isn't in the correct format." has been fixed in v1.10.1. Can you update and try again?
Alright, I've managed to capture the diagnostics during one of these failures: orbstack-diagreport_2025-03-04T16-47-18.021473Z.zip
FWIW, this time my machine's memory was not maxed out. So, I think my previous theory about hitting a memory limit is probably wrong.
I'm seeing this as well on my M4 Pro Mac mini... memory isn't maxed out or anything...
Yep, still happening for me as well. I've had to stop using OrbStack because of it. It's not 100% reliable, but doing a lot of IO through the ClickHouse official containers seems to trigger it more often than anything else.
Same here on an M4 Mac Mini - when doing more I/O it exits the same way. It also exits randomly from time to time, it's very annoying especially for things that need to run continuously. Can someone possibly look into it?
