timers go haywire

Bug #145671 reported by seebs on 2007-09-27
Affects Status Importance Assigned to Milestone

Bug Description

This one's very odd, and I don't have it entirely narrowed down.

The system is an old IBM business desktop running ubuntu 7.04. I disabled X because all I'm using it for is an asterisk box.

The symptom is this: At some point, long after the system boots, timers go haywire. The real-time-clock stays nice and accurate, it stays in sync with every other clock I have, and so on. However, 'sleep 1' can take anywhere from twenty seconds to a minute to complete. This is confirmed identically both by in-system clocks and by a watch external to the system, which agree on how long it takes. The system is NOT under heavy load; it's barely even under a light load. The actual sleep syscall is where all the delay happens; it's not taking a huge time to start the command, it's just that, when an actual request for timer feedback is filed with the kernel, it comes back LUDICROUSLY late.

This affects, so far as I can tell, absolutely everything on the system which uses timers in any way. Shutting down to reboot takes about half an hour once this has struck.

There is no problem at all when the system initially boots. I cannot reproduce the behavior on purpose -- but if I wait a day or so, it always starts happening.

It may be relevant that the system has asterisk on it, and a Digium card which is producing about a thousand interrupts a second. That shouldn't be too many for the system, but for all I know it matters.

Adam Niedling (krychek) wrote :

Is this still occurring in Hardy?

Pedro Villavicencio (pedro) wrote :

We are closing this bug report because it lacks the information we need to investigate the problem, as described in the previous comments. Please reopen it if you can give us the missing information, and don't hesitate to submit bug reports in the future. To reopen the bug report you can click on the current status, under the Status column, and change the Status back to "New". Thanks again!.

To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers