[owncloud-user] Which is the best server configuration for ownCloud?

Diederik de Haas didi.debian at cknow.org
Wed Feb 26 01:02:43 MET 2014

On Wednesday 26 February 2014 00:37:37 Pierre Malard wrote:
> localu at nuage:~# free -h
>              total       used       free     shared    buffers     cached
> Mem:          4,0G       3,7G       279M         0B       395M       1,8G
> -/+ buffers/cache:       1,5G       2,5G
> Swap:         1,0G       7,1M       1,0G
> > One limitation has been that you need more memory than the largest file
> > you sync, so if you sync a 4 gb file, you will need 4gb ram as it all
> > gets loaded into there. This may have been solved with chunked uploads a
> > while back, could someone confirm/deny?

I don't think that's true, even before chuncked uploads.
Note that this is from (quite) a while back, but what I had noticed (back 
then) is that your upload location (usually /tmp) should have enough HDD space 
to store the whole file.
I noticed it when running it in (VBox) virtual machine with only 512MB RAM 
allocated and therefor /tmp was also rather small as Debian allocated (by 
default) 20% of the size of your RAM for /tmp.
You can of course change that by modifying /etc/fstab.

> However, the "free -h" showed before were take without any
> activity. My "owncloud.log" is flat at this time...

Without any activity (and I'm the only user anyway) my 'free -h' stats are:
diederik at cknowsvr01:~$ free -h
             total       used       free     shared    buffers     cached
Mem:          3.9G       247M       3.6G        17M        20M       110M
-/+ buffers/cache:       116M       3.8G
Swap:         4.1G         0B       4.1G

And with me logged in to the web interface:
Mem:          3.9G       409M       3.5G        25M        23M       232M
-/+ buffers/cache:       153M       3.7G
Swap:         4.1G         0B       4.1G

Note that I have very few files etc stored in my/this owncloud instance.

As you can see that is significantly lower then you have and there are a couple 
of more services running on that machine (Debian testing, OC 6.0.1+dfsg-1).
So I recommend 'tearing up' your machine piece by piece to find what's causing 
your memory usage. And by 'tearing up', I mean virtualized vs non-virtualized, 
with NFS and without, SAN vs local disk and synching and no synching.
It would probably also be informative to have (h)top and iotop running while 
performing those tests.

GPG: 0x138E41915C7EFED6
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <http://mailman.owncloud.org/pipermail/user/attachments/20140226/bb84f8f4/attachment.sig>

More information about the User mailing list