[Opensim-dev] Performance issue :S
Ursula MATOVA
ursula.matova at klintcentral.net
Wed Aug 27 05:16:46 UTC 2008
Mariusz,
I'll try to provide as many informations as possible,
but later today because need to go to work.
I'm running Linux Ubuntu 8.0.4TLS 32 bits
Mono from tarball :
Mono JIT compiler version 1.9.1 (tarball)
Copyright (C) 2002-2007 Novell, Inc and Contributors. www.mono-project.com
TLS: __thread
GC: Included Boehm (with typed GC)
SIGSEGV: altstack
Notifications: epoll
Architecture: x86
Disabled: none
Here is the output of "sar -A" hope this could help.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: SAR_20080827.txt.gz
Type: application/x-gzip
Size: 9500 bytes
Desc: not available
URL: <http://opensimulator.org/pipermail/opensim-dev/attachments/20080827/1fb1e91e/attachment-0001.bin>
-------------- next part --------------
Just for your information,
I have added many custom textures to the OpenSim default library and
my ASSETS MySQL table is about 300MB.
I know I have a performance issue with MySQL, but it's ok for now, and
it's not the problem.
The only change was I added too many objects.
These objects are "trees" ( 7 linked prims each ).
Well, must leave now. I will continue to provide informations tonight.
Regards.
Le 27 ao?t 08 ? 01:15, Mariusz Nowostawski a ?crit :
> Hi Ursula,
>
> We have not observed such high CPU utilisation in any older SVN
> revisions in our own testing environment, but unfortunately we've been
> too busy this week to tackle the latest revisions and replicated your
> problems. This may need to wait till late next week, and then I'll be
> able to confirm or pin down the reasons (if any).
> Can you tell me what exact Mono version you use and what's the mono
> configuration (GC used any custom compiling flags, etc). Also, can you
> dump the memory usage together with the CPU usage and I/O so we have a
> slightly more info of what's going on with the system. My guess
> would be
> that there is an excessive amount of paging due to exhaustion of
> physical ram that is causing this - but it may be something else if no
> unusual memory and i/o traces is observed.
>
> cheers
> Mariusz
>
>
>
> Ursula MATOVA wrote:
>> Well,
>>
>> Nobody ??
>>
>> What about performances on "official" grids such as OSGrid ?
>> Sure you run a most powerful server to host all your regions,
>> but, mine has only 25 Regions, about 1500 prims / Regions ...
>> ... no scripts runnning ...
>>
>> And, even in SVN.5992 ...
>> Still have performance issues ( see below ) ...
>>
>> Well, even if my server hasn't the good sizing ( CPU/RAM ) ... It's a
>> goos stress test ... Only 37500 prims on the database.
>>
>> One solution is to get a new server, most powerful ...
>> Second one is to investigate why, without any user logged, and fresh
>> UGAIR startup, I encounter 95%CPU.
>>
>> Regards,
>>
>> Sorry for the noise, ( ans sorry for my poor english )
>>
>> Ursula.
>>
>>
>> Le 25 ao?t 08 ? 08:19, Ursula MATOVA a ?crit :
>>
>>
>>> Hi All,
>>>
>>> Just a very quick message.
>>>
>>> I have loaded my Regions with about 1500prims on Each.
>>>
>>> Everything runs ok except my CPU LOAD is over 90% :
>>>
>>> TIME CPU %user %nice %system %iowait %steal
>>> %idle
>>> 02:05:01 all 94,88 0,00 0,92 0,02
>>> 0,00 4,18
>>> 02:15:01 all 94,80 0,00 0,84 0,00
>>> 0,00 4,36
>>> 02:25:02 all 94,40 0,00 0,84 0,00
>>> 0,00 4,76
>>> 02:35:02 all 94,98 0,00 0,82 0,00
>>> 0,00 4,19
>>> 02:45:01 all 94,62 0,00 0,80 0,00
>>> 0,00 4,57
>>> 02:55:02 all 94,64 0,00 0,84 0,00
>>> 0,00 4,52
>>> 03:05:01 all 94,80 0,00 0,83 0,00
>>> 0,00 4,38
>>> 03:15:01 all 94,31 0,00 0,86 0,00
>>> 0,00 4,83
>>> 03:25:01 all 94,96 0,00 0,81 0,00
>>> 0,00 4,23
>>> 03:35:01 all 94,18 0,00 0,85 0,00
>>> 0,00 4,97
>>> 03:45:02 all 94,84 0,00 0,81 0,00
>>> 0,00 4,35
>>> 03:55:01 all 94,57 0,00 0,82 0,00
>>> 0,00 4,61
>>> 04:05:01 all 94,56 0,00 0,81 0,00
>>> 0,00 4,63
>>> 04:15:01 all 94,60 0,00 0,82 0,00
>>> 0,00 4,57
>>> 04:25:01 all 94,39 0,00 0,84 0,00
>>> 0,00 4,77
>>> 04:35:02 all 94,91 0,00 0,83 0,00
>>> 0,00 4,26
>>> 04:45:02 all 94,72 0,00 0,78 0,00
>>> 0,00 4,50
>>> 04:55:01 all 94,59 0,00 0,87 0,00
>>> 0,00 4,54
>>> 05:05:01 all 94,47 0,00 0,82 0,00
>>> 0,00 4,72
>>> 05:15:02 all 94,49 0,00 0,83 0,00
>>> 0,00 4,68
>>> 05:25:01 all 94,52 0,00 0,81 0,00
>>> 0,00 4,67
>>> 05:35:01 all 94,70 0,00 0,83 0,00
>>> 0,00 4,48
>>> 05:45:01 all 94,28 0,00 0,87 0,00
>>> 0,00 4,86
>>> 05:55:01 all 94,52 0,00 0,85 0,00
>>> 0,00 4,64
>>> 06:05:01 all 94,76 0,00 0,84 0,00
>>> 0,00 4,40
>>> 06:15:01 all 94,48 0,00 0,88 0,00
>>> 0,00 4,64
>>> 06:25:01 all 94,85 0,00 0,82 0,00
>>> 0,00 4,34
>>> 06:35:01 all 94,55 0,00 0,88 0,01
>>> 0,00 4,56
>>> 06:45:01 all 95,06 0,00 1,18 0,05
>>> 0,00 3,72
>>> 06:55:01 all 95,83 0,00 1,08 0,00
>>> 0,00 3,09
>>> 07:05:02 all 94,46 0,00 0,87 0,00
>>> 0,00 4,66
>>> 07:15:01 all 94,83 0,00 0,85 0,00
>>> 0,00 4,32
>>> 07:25:01 all 94,14 0,00 0,87 0,00
>>> 0,00 4,99
>>> 07:35:01 all 94,83 0,00 0,83 0,00
>>> 0,00 4,35
>>> 07:45:01 all 94,86 0,00 0,78 0,02
>>> 0,00 4,34
>>> 07:55:02 all 94,38 0,00 0,90 0,00
>>> 0,00 4,72
>>>
>>> I run OpenSim in Grid mode, SVN.5969, on Linux 32bits Ubuntu
>>> 8.0.4TLS.
>>> The server is based on a "Via C7/2GHz" CPU with 1GB RAM.
>>> The Grid has 25 Regions.
>>>
>>> Before I encountered this problem ( previous SVN releases ), the
>>> only
>>> difference is the number of prims ( Globaly : 37500 prims ).
>>>
>>> Are you experiencing the same ?
>>>
>>> I'll make some more tests tonight ( time to go to work :D )
>>>
>>> Regards,
>>> Ursula.
>>>
>>>
>>> _______________________________________________
>>> Opensim-dev mailing list
>>> Opensim-dev at lists.berlios.de
>>> https://lists.berlios.de/mailman/listinfo/opensim-dev
>>>
>>>
>>
>> _______________________________________________
>> Opensim-dev mailing list
>> Opensim-dev at lists.berlios.de
>> https://lists.berlios.de/mailman/listinfo/opensim-dev
>>
>
> _______________________________________________
> Opensim-dev mailing list
> Opensim-dev at lists.berlios.de
> https://lists.berlios.de/mailman/listinfo/opensim-dev
>
More information about the Opensim-dev
mailing list