Hi,

I'm having serious problems with my NAS. I've bought it to serve several purposes. One of the major being a centralized backup. I've got a Synology diskstation DS512j and two 4 TB harddisks. The idea was that one harddrive, /volume1/ should contain "stuff". Photos, music, movies, git, calendar, contacts, notes, documents I need access to when I'm out, etc. The other drive, /volume2 (ext4), was going to be backup.

I cannot get the backup system to work reliably. I've already got backup set up on my computer using rsnapshot. I expected I could simply rsync the content of my existing backup to my diskstation and then, using NFS, mount the diskstation to where I take my backup today.

That does not work, however, and it behaves very strangely.

OK. First things first, diskstation does not allow you to change the UID of the users (For whatever cleaver reason. Somebody made a mistake when designing the UI, and assumed everybody wanted the same UIDs they preferred, and now the UI doesn't work if one change the UIDs). Anyway, I'm not going to adapt my four computers to the liking of Synology. For that I'm to stubborn. So I use bindfs to map my UIDs to the diskstation UIDs. I doubt this is why it fails, but I guess I could try without and see what happens.

After rsyncing my backup, which takes around 20 hours with my 100Mb/s connection, I tried running rsnapshot. What happened was very strange. For some reason daily.6 had the most recent date, and it contained only a few folders and no files. The backup was now 30 GB bigger than my local backup. (I think it was actually bigger than the local backup right after rsync)

I thought something strange happened, so I deleted the folder. The size of my backup was now 1/8th less than before, and more comparable to my local backup. Since I didn't want to start all over again, I wanted to rsync (rsync -a -H --delete /backup/ /mnt/backup/) to make sure the NAS backup was up to speed with my local backup. Now the backup on my NAS is 25 (and increasing) bigger than my local backup, so it's totally failing in that respect.

Also, while I'm taking backup, periodically my diskstation stalls because nfsd is hogging the CPU driving the transfer rate down by over 95.

What's going on?

Edit:
Ok, I was working on this yesterday, around midnight, after a couple beers. After rsyncing the diskstation I took two backups. One to /backup and one to /mnt/backup. At least that's what I thought. Apparently I took the backup to /etc/backup. Before actually migrating to diskstation I want to confirm that rsnapshot works, and also do a md5sum check of both trees. That's why I for now take two backups.

So rsnapshot can, for now, be disregarded. I still don't understand why rsync fails so misserably though.

An update to this issue.

I disabled bindfs, but no increase in performance was observed. I have to experiment more with rsync to see why it will not sync correctly. I supposed it has something to do with hardlinks being lost and rsync will produce actual copies which drives the space requirements up.

rsnapshot also has a strange behavior. It only works when I haven't had any beers.. :P

Seriously though. It seems to work, though nfs is still _extremely_ slow. For comparison, it took 2 minutes to do do a backup to an internal HD. The same backup took over 2 hours through NFS.
The network link was not the problem, as it only utilized ~10Kb/s. The problem is the nfsd hogging the CPU.

NFS reportedly have 2-3 times higher performance when mounted async compared to sync. I will try this tomorrow and see if it has any influence.
I might try sshfs also and see how it perform the day after tomorrow.

AFter that I'm out of ideas.