[OpenIndiana-discuss] Memory drains away....

Fri Nov 4 13:22:47 UTC 2011

I'm wondering if this memory drain may be due to zpool caching (L2ARC). 
You could try and see if it helps of you set a limit on this cache. The 
way that I understand it zpool/zfs uses quite a bit of RAM for caching.

Robin.

On 2011-11-04 14:13, Daniel Kjar wrote:
> pmap seems to give me exactly what top tells me.
>
> There just doesn't look like 32gb of stuff running (more like 5gb 
> max)...  although I have to laugh at how much memory Firefox seems to 
> need....
>
>  9409 eagle3     17  59    0  765M  739M sleep    3:24  0.37% VirtualBox
>   9413 eagle3     21  39    0  291M  129M sleep    1:54  0.51% firefox
>   2589 eagle2     21  59    0  245M  150M sleep    0:34  0.01% java
>   5357 eagle2      1  59    0  112M   58M sleep    0:01  0.00% nautilus
>   4179 eagle3      1  58    0  107M   98M sleep    1:03  0.03% Xnewt
>   4470 eagle3      1  49    0  106M   39M sleep    0:01  0.00% nautilus
>   3423 gdm         1  59    0  102M   35M sleep    0:00  0.00% 
> gnome-settings-
>   5063 eagle2      1  59    0   98M   90M sleep    0:33  0.00% Xnewt
>   3430 gdm         1  59    0   91M   26M sleep    0:57  0.01% 
> gdm-simple-gree
>   4480 eagle3      1  59    0   91M   21M sleep    0:03  0.00% 
> isapython2.6
>   4467 eagle3      1  59    0   90M   24M sleep    0:01  0.00% 
> gnome-panel
>   5354 eagle2      1  59    0   90M   26M sleep    0:01  0.00% 
> gnome-panel
>   4499 eagle3      1  59    0   86M   19M sleep    0:21  0.01% 
> mixer_applet2
>   5388 eagle2      1  59    0   86M   19M sleep    0:15  0.01% 
> mixer_applet2
>   4462 eagle3      1  59    0   86M   18M sleep    0:00  0.00% 
> gnome-settings-
>   4486 eagle3      1  49    0   86M   18M sleep    0:00  0.00% 
> trashapplet
>   3429 gdm         1  59    0   85M   18M sleep    0:01  0.00% 
> gnome-power-man
>   5371 eagle2      1  59    0   84M   17M sleep    0:00  0.00% 
> wnck-applet
>   4485 eagle3      1  59    0   84M   17M sleep    0:01  0.00% 
> wnck-applet
>   4466 eagle3      1  59    0   84M   17M sleep    0:00  0.00% metacity
>   5360 eagle2      1  59    0   83M   17M sleep    0:01  0.00% 
> gnome-power-man
>   4471 eagle3      1  59    0   83M   16M sleep    0:01  0.00% 
> gnome-power-man
>   5373 eagle2      1  59    0   79M   13M sleep    0:00  0.00% 
> trashapplet
>   3428 gdm         1  59    0   78M   13M sleep    0:00  0.00% metacity
>    637 mysql       9  59    0   53M   19M sleep    0:19  0.01% mysqld
>   2452 root       23  59  -10   52M   26M sleep    0:45  0.02% java
>   5362 eagle2      1  12   19   49M   33M sleep    0:02  0.00% 
> updatemanagerno
>   4473 eagle3      1  12   19   49M   33M sleep    0:02  0.00% 
> updatemanagerno
>   3367 root        4  59    0   47M   36M sleep   13:58  0.17% Xorg
>   3332 root        1  59    0   39M   20M sleep    0:20  0.01% Xnewt
>
>
> On 11/ 4/11 09:01 AM, Michael Schuster wrote:
>> On Fri, Nov 4, 2011 at 13:56, Daniel Kjar<dkjar at elmira.edu>  wrote:
>>> Just serving sunray sessions but this happens whether anyone is 
>>> logged on
>>> (none last night) or not.  If somebody fires up something like a 
>>> virtual
>>> machine it just speeds up the process.
>> hmm ... (haven't done this in a while ...) I'd suggest you
>> periodically (once every 10', perhaps)  'pmap' all processes and see
>> which grow most, and then look at those in detail (perhaps using pmap
>> -x, or starting those with LD_PRELOAD=/usr/lib/libumem.so ...)
>>
>> I'm sure other people have better ideas.
>>
>> HTH
>> Michael
>>> On 11/ 4/11 08:53 AM, Michael Schuster wrote:
>>>> Hi,
>>>>
>>>> the obvious question to ask: what's the machine doing?
>>>>
>>>> Michael
>>>>
>>>> On Fri, Nov 4, 2011 at 13:49, Daniel Kjar<dkjar at elmira.edu>    wrote:
>>>>> I have always had this problem and it has never been resolved.  Does
>>>>> anyone
>>>>> else see  this and is there a cure?
>>>>>
>>>>> I have a v40z (although all of my boxes do it) 32gb ram 4 dual 
>>>>> core 885s.
>>>>>   When you restart the server it shows about 27 gb free.  Over 12 
>>>>> hours
>>>>> this
>>>>> slowly degrades to about 1gb free and I start to notice a slow down.
>>>>>   Once
>>>>> this drops down to 400mb stuff starts to really drag.  Setting the 
>>>>> max
>>>>> zfs
>>>>> arc use does nothing.  This is using TOP.
>>>>>
>>>>> If you look at the gnome performance monitor it will say you are only
>>>>> using
>>>>> 2gb of ram yet there is degradation in performance (I assume for slow
>>>>> memory
>>>>> clearing and reuse).
>>>>>
>>>>> CPU states: 94.3% idle,  1.3% user,  4.4% kernel,  0.0% iowait,  0.0%
>>>>> swap
>>>>> Kernel: 7755 ctxsw, 139 trap, 13034 intr, 22379 syscall, 120 flt, 136
>>>>> pgin
>>>>> Memory: 32G phys mem, 988M free mem, 16G total swap, 16G free swap
>>>>>
>>>>> and this dropped below 1gb while I was typing this message.
>>>>>
>>>>>
>>>>> -- 
>>>>> Dr. Daniel Kjar
>>>>> Assistant Professor of Biology
>>>>> Division of Mathematics and Natural Sciences
>>>>> Elmira College
>>>>> 1 Park Place
>>>>> Elmira, NY 14901
>>>>> 607-735-1826
>>>>> http://faculty.elmira.edu/dkjar
>>>>>
>>>>> "...humans send their young men to war; ants send their old ladies"
>>>>>         -E. O. Wilson
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> OpenIndiana-discuss mailing list
>>>>> OpenIndiana-discuss at openindiana.org
>>>>> http://openindiana.org/mailman/listinfo/openindiana-discuss
>>>>>
>>>>
>>> -- 
>>> Dr. Daniel Kjar
>>> Assistant Professor of Biology
>>> Division of Mathematics and Natural Sciences
>>> Elmira College
>>> 1 Park Place
>>> Elmira, NY 14901
>>> 607-735-1826
>>> http://faculty.elmira.edu/dkjar
>>>
>>> "...humans send their young men to war; ants send their old ladies"
>>>         -E. O. Wilson
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> OpenIndiana-discuss mailing list
>>> OpenIndiana-discuss at openindiana.org
>>> http://openindiana.org/mailman/listinfo/openindiana-discuss
>>>
>>
>>
>