[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Kernel lockup (kjournald?)



Hi,

On Wed, 2003-04-02 at 22:55, Kevin Clark wrote:
> I am getting an odd situation when backing up a number of ext3 filesystems
> and was wondering if it could be caused by journalling.  Over the space of
> a minute the load average will jump from 2 to over 40 and the system will
> be unresponsive for anywhere from 8 to 25 minutes.  I am going to be trying
> a number of things, but was wondering if anyone could see the reason for
> the high load given the following statistics collected after the system
> started to semi-respond.

No, nothing that I can see.  I'd check the system logs for signs of
problems in the IO subsystem, though.

> It is not clear what is being reported by the "load average" indicator
> under the Linux operating system... one can assume that it is including
> processes blocking on IO.

That's right --- it's the total number of tasks in either R or D state. 
And in your trace, there are only 5 such tasks, so the trace has
obviously missed the point of massive system load.

Cheers,
 Stephen





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]