On Fri, Apr 13th, 2012 at 3:41 PM, Mark Post wrote:
> /proc/*/smaps exists in SLES10, but there's no "Swap:" fields in them.
Arrgggh. That'll certainly get a bunch of zeroes out of that script.
Note to self, _never_ presume nuthin ...
Thanks Mark.
---
It appears my script was not good for a sles10 system as I was informed
by IBM tonight that "Determining swap space used by a particular process
from smaps doesn't
appear to be available until SLES11 kernels"
So I'm still looking for a way to see this but that isn't the way.
Phil
>>> On 4/12/2012 at 10:18 AM, PHILIP TULLY wrote:
> From the looks of this system, I suspect an overnight backup process
> utilizing the full memory allocation and populating swap. That process
> then finishes but swap doesn't' clear because of it's "lazy" attributes.
If you're collecting SAR d
Rob/Shane,
I appreciate your overnight assistance, I did modify the script to look
at all pids, with the same result.
I do realize that linux memory mgmt is lazy (delayed cache writes) and
that is why I was wondering if I can't find which current pid
has swap allocations, is there a way to run t
On Thu, Apr 12th, 2012 at 8:37 PM, Rob van der Heij wrote:
> With 22 processes having this mapped, I would count it as 22 times 8
> kB while it really is just 8 kB on swap? And how come part of this is
> private when it's read-only?
Note the last sentence of my previous post. That applies (partic
On Thu, Apr 12, 2012 at 11:02 AM, Shane G wrote:
> No, by lazy in this context I meant that freed memory (pages) are not
> immediately moved to the free list. This even extends to task termination.
> If memory pressure ramps up sufficiently, kswapd will get kicked to balance
> out the trees. Coul
No, by lazy in this context I meant that freed memory (pages) are not
immediately moved to the free list. This even extends to task termination.
If memory pressure ramps up sufficiently, kswapd will get kicked to balance
out the trees. Could take a while - like forever.
In addition to what Rob men
On Thu, Apr 12, 2012 at 4:09 AM, PHILIP TULLY wrote:
> I was using the following to show how much swap space was being used it
> is either not working or there is nothing allocated to swap.
>
> for pid in `ps -ef|grep ora| awk '{print $2}'`; do echo -n "Pid: $pid
> "; cat /proc/$pid/smaps |grep
- obvious first step would be to check the entire system, rather than a
subset you obviously think is the cause.
- smaps should be believed.
- Linux uses lazy (memory) allocation. This includes de-allocation. And swap.
Hence the various tools that simply read meminfo should be treated with
(extreme
I have a couple servers which show 100% swap space used but when I look
I can't find which pids are have pages in swap.
free -m
total used free sharedbuffers
cached
Mem: 12061 11410650 0152
6425
-/+ buffers/cache:
10 matches
Mail list logo
