[Pan-users] Re: Strange 0.131 behaviour on openSUSE 10.2

[walt]

On Mon, 30 Jul 2007 23:04:49 +0000, Jim Henderson wrote:

> Just did a fresh install of openSUSE 10.2 (hard drive crashed on me),
> and restored a backup of my home directory data from 10.1.
> 
> I noticed that 0.131 on 10.2 tends to have long periods of burying the
> processor when toggling the view from "Match Only Unread Articles".
> Given that I was running a binary from 10.1, I wiped it out and
> recompiled natively on 10.2, but I still see this behaviour.
> 
> Typically, the processor pegs at >99% for anywhere from 5 to 15 minutes,
> with pan hogging the CPU; not so bad that other applications stop
> responding, but pan is completely unusable in this state (its GUI stops
> refreshing even).
> 
> strace during the hang shows the following information:
> 
> --- snip ---
> 
> read(13,
> writev(13,
> poll([{fd=6,
> {fd=11,
> {fd=10, 
> {fd=12,
> {fd=13,
> read(13,
> --- snip ---
> 
> This loops over and over for a very long time, and very quickly.
> 
> Anyone have any ideas as to what's happening here?

Although Duncan's machine is mucho macho compared to mine ;o) I
still see instantaneous response when toggling between All and
Unread articles, so something must be very wrong at your end.

I agree that my first step would be to rename my .pan2 directory
and let pan start fresh with a new one.

If that doesn't fix it, then I'd be curious to know which files
pan is thrashing during those minutes while you wait.  The 'fd'
can often be matched with a filename in the strace output if you
look far enough back in the output to find where the file was first 
opened.

Also, is your swapfile being used during these episodes, i.e disk
thrashing as well as file thrashing?