Re: IO latency - a special case

[Trenton D. Adams]

Sorry I didn't reply earlier.  I got busy with some server troubles and such.

On Thu, Jun 11, 2009 at 1:11 PM, Theodore Tso<tytso@xxxxxxx> wrote:
> On Sat, Jun 06, 2009 at 09:14:43PM -0600, Trenton D. Adams wrote:
>> I'm back into doing some Java development, and it's happening again.
>> I'm running 2.6.30-rc7, and I did a SysReq+T, I hope this is helpful.
>> I've attached a txt.gz file, because it's 367K to my
>> /var/log/messages.
>>
>> The important stuff, I believe, is the Java related stuff.  As a quick
>> refresher to this problem, it appears to be related to how my Java IDE
>> saves the file, and perhaps something doesn't release the futex right
>> away?
>
> Have you checked your memory usage via SysReq-m, or /proc/meminfo or
> simply using "free".  I'm seeing a lot of processes in do_page_fault,
> which is interesting, especially in the non-Java processes.

I wasn't even aware of a SysReq-m.  I'll try that next time.  <rant>I
do wish the Macbook had a hard drive activity light *rolls
eyes*.</rant>

>
> Back in April, your vmstat output didn't show much swapping activity,
> but maybe there was more paging activity that didn't show up there?

Not sure I understand.  Paging without swapping?  I thought they were
one in the same, please explain; it's been a long time since I studied
kernel related theory.

>
>> I've also seen these symptoms previously...
>>
>> 1. Problem arises, I Ctrl-C the process when I notice the system
>> getting really sluggish.
>> 2. Start process again, and it locks up the machine for about an hour
>
> What sort of process where you typicalling trying to kill using ^C?

It's maven, which spawns javac (java compiler), as well as does
various other tasks such as making connections to the Internet, local
directories, etc.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/