Has anyone set up Nextcloud with a Hetzner Storagebox cifs mount as primary storage?

[msbt]

@andreasdueren I've mounted it via volumes after getting rid of fstab

[marcusquinn]

It looks like Hetzner are now pretty-much doing this exact same thing for their "Storage Share" solution:

• https://www.hetzner.com/storage/storage-share

@girish I think this needs to be a core feature of Cloudron, given the cost-efficiency, scaleability, and additional backup snapshots this offers for very little €s compared to things like GDrive & Dropbox per-user pricing.

[jcgonnard]

@msbt
Hi

I got a similar configuration with a CPX21 instance connected to a BX21 Storagebox.
While transfer speed for large files is fine, transfer speed for large amount of small files is horrible:

root@nextcloud:~# for f in {1..2000}; do mktemp ; done 
/tmp/tmp.qabMyUDO5N
/tmp/tmp.QPuk8CcXSB
/tmp/tmp.NhErmerWFl
/tmp/tmp.CtkY3jvTbL
[...]

root@nextcloud:~# time cp /tmp/tmp.* /backups/test/
real	4m20.998s
user	0m0.061s
sys	0m0.565s

root@nextcloud:~# time rm -fr /backups/test/*
real	1m36.682s
user	0m0.050s
sys	0m0.196s

Are you experiencing the same behaviour ?

[robi]

@jcgonnard many factors influence the availability of IO on the system.

Shared systems get busy.

See what is available with dd:
dd if=/dev/zero of=/tmp/test1.img bs=1G count=1 oflag=dsync ; rm /tmp/test1.img

[jcgonnard]

@jcgonnard

SSHFS:

root@nextcloud:~# dd if=/dev/zero of=/backups/test.img bs=1G count=1 oflag=dsync
1+0 records in
1+0 records out
1073741824 bytes (1.1 GB, 1.0 GiB) copied, 50.1201 s, 21.4 MB/s
root@nextcloud:~# dd if=/dev/zero of=/backups/test.img bs=4M count=250 oflag=dsync
250+0 records in
250+0 records out
1048576000 bytes (1.0 GB, 1000 MiB) copied, 78.6863 s, 13.3 MB/s
root@nextcloud:~# dd if=/dev/zero of=/backups/test.img bs=1 count=1000 oflag=dsync
1000+0 records in
1000+0 records out
1000 bytes (1.0 kB) copied, 70.862 s, 0.0 kB/s

LOCAL:

root@nextcloud:~# dd if=/dev/zero of=/tmp/test.img bs=1G count=1 oflag=dsync
1+0 records in
1+0 records out
1073741824 bytes (1.1 GB, 1.0 GiB) copied, 0.863714 s, 1.2 GB/s
root@nextcloud:~# dd if=/dev/zero of=/tmp/test.img bs=4M count=250 oflag=dsync
250+0 records in
250+0 records out
1048576000 bytes (1.0 GB, 1000 MiB) copied, 1.31499 s, 797 MB/s
root@nextcloud:~# dd if=/dev/zero of=/tmp/test.img bs=1 count=1000 oflag=dsync
1000+0 records in
1000+0 records out
1000 bytes (1.0 kB) copied, 0.500667 s, 2.0 kB/s

As I said, transfer speed for medium/big files is fine enough, the problem is for small files.
I just asked if some of you experienced the same issue with the same configuration (hetzner VM + storagebox) and if there is a way to improve performances or if I need to find something else that would match my needs

[robi]

@jcgonnard that is pretty good performance for SSHFS.

You'll have to switch to a different mount option to increase small file performance, unless you can get all your small files to stream over via tar/zip/etc.

[MooCloud_Matt]

@marcusquinn
Hetzner don't use remote storage use remote clustered DB + local storage.
So it's not possible to compare the 2 installs cloudron + storagebox and hetzner storageshare, performance will be a lot better on storageshare, in order of magnitude especially on small files.

[msbt]

@jcgonnard I can't really tell, it's always running in the background and I haven't encountered any issues so far

[MooCloud_Matt]

@msbt
the issue could start with a lot of concurrent users and a lot of small changes to file, due to SMB not working at FS lvl but at file level/block of data.
IF hetzner would offer NFS, that could be a solution or better even if there would be support for iscsi.

[JOduMonT]

I would recommend you to use WebDAV instead of CIFS
it will be faster as WebDAV is more modern and appropriate for extended network (non LAN).