Exposing Zone Sleeps

[jhell]

On 05/26/2010 17:31, jhell wrote:
> On 05/26/2010 13:29, Sean Bruno wrote:
>> On Tue, 2010-05-25 at 07:47 -0700, Sean Bruno wrote:
>>>> Hi Sean,
>>>>
>>>> Nice work on this. I applied this to stable/8 r208530 and I am in the
>>>> process of compiling the kernel right now. Everything else has built &
>>>> runs as expected "i386". Attached is the adjusted patch which was one
>>>> modification to the line number for uz_sleeps in sys/vm/uma_int.h.
>>>>
>>>> 8 files changed, 106 insertions(+), 7 deletions(-)
>>>>
>>>> For those wishing to apply this patch and test for them self:
>>>>
>>>> cd /usr/src
>>>> patch </path/to/sleep_stat_stable8_r208530.diff
>>>> cd /usr/src/include
>>>> make obj && make depend && make includes && make install
>>>> cd /usr/src/lib/libmemstat
>>>> make obj && make depend && make includes && make install
>>>> cd /usr/src/usr.bin/vmstat
>>>> make obj && make depend && make install
>>>> cd /usr/src
>>>> make kernel KERNCONF=YOUR_KERN_CONF
>>>> reboot
>>>>
>>>> Can't wait to see some results from this & I will report back with
>>>> either negative results of the build & run or positive results from the
>>>> stats collected.
>>>>
>>>> If there is anything needed feel free to let me know and I will do what
>>>> is possible ASAP.
>>>>
>>>> Thanks again,
>>>>
>>>> - -- 
>>>>
>>>>  jhell
>>>
>>> Excellent.  Please check the output of vmstat -z and the appropriate
>>> sysctl.  I changed the display a bit to keep it from wrapping on a
>>> standard terminal.
>>>
>>> Sean
>>>
>>> P.S.  My intention it to MFC this to all releases.
>>>
>>
>> I do have a concern related to the removal of an #ifdef DDB in this
>> patch.  Any comments?
>>
>> Sean
> 
> This was in your original patch sent to the list. I questioned it too
> but as far as testing it goes it has caused no harm that I can see here
> but I will add those back in along with the improvements from Garrett,
> then regenerate the patch and send it back to the list.
> 
> Regards,
> 

Following up on this, it has been ~ 1 -> 2 weeks with this patch on 4
machines I have had in question and have not noticed any negative
effects or gains from what this patch provides. On a stable/8 i386
system there were absolutely no SLEEP stat bumps recorded at all.

On a second note one negative impact that I had observed is the negative
ability to upgrade a stable/7 system to a stable/8 system with this
patch applied due to mismatched symbols or other related symptoms that
would suggest that the patch would also have to be applied to the older
system and compiled before the upgrade.  This all had more to do with
'vm_uma_zone' being undeclared.

Although this is a nice feature I do not see any benefits for applying
it to a stable/7 or stable/8 systems as those are already clean.  But
applying this to HEAD for debugging of sleep issues that may creep into
the code would be useful as a type of alert until the next release at
which it should be turned off and ejected from the tree unless it was
separated into the src/tools/tools directory as a debugging tool.

With all luck & kind regards,

-- 

 jhell