Error: Timeout: agent connection was never initiated (null) (URL (null)) #2794
Replies: 3 comments 4 replies
-
Yeah this is an unreasonably long delay to start the sparkle-progress / Updater.app. I have already tried hinting to bump priority of these launchd jobs and set generous timeouts in #2162 I suspect this particular machine has some systematic / slowness issues(?). Is there anything behaving slowly with launching apps? For example I've seen unresponsiveness in launching tasks when one of my machines was very low on disk space. Since the same app binary works on other machines I think we can rule out issues due to incorrect integration of the framework. Moving to discussion for now unless we find there's anything actionable on Sparkle side. I've not seen any other reports for this and this particular code area has not changed in a long while. Also betting the underlying issue is outside of Sparkle. Chances are a reboot could fix it but then you may not be able to diagnose issue further. |
Beta Was this translation helpful? Give feedback.
-
|
On second thought
This may not be right for this service to be on-demand-only. I'll look later at the launchd keys we're passing.. |
Beta Was this translation helpful? Give feedback.
-
|
@Tom-and-Jerry-gif If you're able to try out #2795 | https://github.com/sparkle-project/Sparkle/actions/runs/18961689872 (there's a Sparkle-distribution-xcode16.4.tar.xz attached) by chance. In the bad case we do not want to see |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for your answer! I'll give it a try |
Beta Was this translation helpful? Give feedback.
-
|
I have made a new discovery regarding the behavior of |
Beta Was this translation helpful? Give feedback.
-
|
Yeah that is kind of expected. probeInstallerInProgressForHostBundleIdentifier can test whether the launchd service is there or not yet at the beginning of the update cycle. Whereas registering/launching the services is more towards the end of the update cycle (so we want the service to be launched without reliance on this probe message). In #2795 I'm trying to make it more "launch when registered" rather than on-demand to begin with. Did you test with those changes above (they are not available in package managers yet)? That said, if it was truly/fully on-demand today, Sparkle wouldn't work.. so I suspect it is only on-demand in some specific cases (system under pressure?). Your success log on your other machine confirms this. |
Beta Was this translation helpful? Give feedback.
-
|
Yeah I have replaced the original Sparkle library with the package mentioned above #2795 yesterday. The timeout issue has not recurred. Thank you again for your reply. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
When updating my app using Sparkle, I'm encountering a timeout error on one specific macOS 26 machine. The error is:
Error: Timeout: agent connection was never initiated (null) (URL (null))On other machines (same Sparkle package, identical update steps), the update completes without issue.
The update error log is as follows:
The normal update log is as follows:
I investigated the issue using log show, and found that:
On the affected machine, after submitting the xxx-sparkle-progress job to launchd, the process did not start immediately. There was a delay of more than 1 minute before the actual spawn and subsequent agent connection, resulting in a timeout.
On the working machine, [bundleId]--sparkle-progress and xxx--sparkle-updater are both launched almost simultaneously. Both are triggered promptly (semaphore acquired), and the update works correctly
Questions
Is there any way to further diagnose or locate the cause of the delayed pending spawn of xxx-sparkle-progress by launchd? What could cause such a delay, given that the update works normally on other machines with the same environment?
Additional context
the timeout only occur on some macOS 26 installations, not universally.
Beta Was this translation helpful? Give feedback.
All reactions