Well, I thought I would respond back to this so people don't waste their 
time.

I'd recommend not using s3ql on top of acd_cli, it's too unstable. I found 
myself with a crashed file system and having to run fsck everyday. My 
reasons for wanting to use s3ql were:

1. The chunked storage format I thought would be better for doing things 
like resuming movies in the middle on my plex server
2. The encryption it provides makes me feel like a sneaky secrete agent.

Instead, I recommend you just upload your movies normally and mount using 
acd_cli.

For #1, chunking the files doesn't seem necessary. I don't know what black 
magic the acd_cli mount is doing, but my movies resume from the middle just 
fine, and I'd actually say that they play a little faster, s3ql seems to 
add some overhead.

I do miss out on the encryption, but I really don't think amazon cares 
about what I upload. The other thing you need to consider with encryption 
and s3ql in general is that you're locking yourself in to a particular 
technology. Right now I'm running plex on linux, but if I ever decide to 
buy a windows license so I can watch netflix on it, I'd have to re-download 
all my movies then re-upload them in some new format. Finally, if 
encryption is really important to you, there's a pull request for 
supporting encryptfs with acd_cli that seems a lot more mature then the 
pull request here for supporting s3ql.

Some more details on what I ended up doing:

1. I am still using union-fs. I keep everything on the local disk until I 
need to make some room, then I archive stuff to acd. It's sorta like a poor 
mans version of the disk cache s3ql provides.
2. rclone is amazing. Instead of using acd_cli to upload which I find to be 
a little slow, I use rclone. I actually got banned for a little bit from my 
vps until I throttled it because I was uploading at 50-80 MBps (capital B 
is intentional!), which was most of their 1Gbps pipe. Now I upload at 
20MBps consistently. Rclone also seems a little bit more stable then 
acd_cli, although their mount stuff is not currently as good.

Anyways, the caching + encryption stuff that s3ql does is cool and I might 
revisit if it ever gets official support. It would be nice if it did 
because then I could remove all my unionfs and acd_cli stuff and just use 
s3ql. But what I have right now is good enough for my needs. I've had 1 
file system crash in the last 2 weeks, and recovery from that is a lot 
faster since I don't have to run fsck.

I know a lot of people are trying to get this combo working, so hopefully 
my post helps.

On Wednesday, October 26, 2016 at 3:04:25 PM UTC-4, Nikolaus Rath wrote:
>
> On Oct 26 2016, Jonas Lippuner <[email protected] <javascript:>> wrote: 
> > I'm using a similar setup. Mount ACD locally with acd_cli and then use 
> an 
> > s3ql volume inside the mounted ACD path. The reason for using ACD rather 
> > than S3 is very simple: cost. 
>
> Well, that explains it. You get what you pay for :-). 
>
> [...] 
> > Here's the backtrace (from mount.log) of the error I usually see: 
> [...] 
>
> > 
> "/usr/local/lib/python3.5/dist-packages/s3ql-2.20-py3.5-linux-x86_64.egg/s3ql/backends/local.py",
>  
>
> > line 304, in __init__ 
> >     self.fh = open(name, 'wb', buffering=0) 
> > OSError: [Errno 14] Bad address: 
> > 
> '/home/jlippuner/cloud_storage/ACD/remote/s3ql_backup/s3ql_data_/698/s3ql_data_69888#7693-140501390976768.tmp'
>  
>
>
> Yes, that's ACD being buggy. 
>
> > I wonder how hard it would be to add a native ACD interface to S3QL... 
>
> Some people are trying, see https://github.com/s3ql/s3ql/pull/7 
>
>
> Best, 
> -Nikolaus 
> -- 
> GPG encrypted emails preferred. Key id: 0xD113FCAC3C4E599F 
> Fingerprint: ED31 791B 2C5C 1613 AF38 8B8A D113 FCAC 3C4E 599F 
>
>              »Time flies like an arrow, fruit flies like a Banana.« 
>

-- 
You received this message because you are subscribed to the Google Groups 
"s3ql" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to