Resume recursive scp transfer (with rsync?)

I was transferring several thousand files each ~1MB via scp and my connection was broken after the first 2k files or so. I wanted to know if there was a way to resume the recursive transfer w/o starting over. Something like

$ scp -r me@host.com:/datafiles/ ./
... Happy Transfer ...
...     BREAK!     ...
$ rsync -P me@host.com:/datafiles/ ./
... Continue transf...

The problem is I can't seem to get the syntax correct if it is possible. Can anyone shed some light on if/how it can be done?

PS. If you specify the slash after "datafiles" in the rsync line, does that transfer the directory or its contents? I saw conflicting comments when I googled.

physicsmichael

Posted 2010-06-01T15:52:16.867

Reputation: 918

Answers

if you are rsyncing from a local machine to a remote host, this would work:

rsync -avzl -e ssh /directory/with/files/ you@host.com:/new/directory/

kibitzer

Posted 2010-06-01T15:52:16.867

Reputation: 216

Thanks. I didn't think this was working because it would begin listing all my files but I didn't realize that it was showing the "file transfer" which just mean noticing the file was up to date. Right were my transfer left off it picked up with real (slower) transfers. – physicsmichael – 2010-06-01T16:23:09.383

Hmm, for me ssh -i akey.pem is required and I can't get the -i part to work with these commands. – isomorphismes – 2012-09-24T03:26:44.500

2If you are trying to resume a large file, this options may be useful: --partial --inplace --no-whole-file. – Rafael Xavier – 2013-03-11T18:59:07.363

@kibitzer - once it showed you the file (on remote server), how did you start the transfer again? Mine just build the list file and shows me the file and how much was uploaded. But it just hangs with no option to continue or progress output. Has the transfer started again or do I need to do something to start it? – masterninja01 – 2014-03-23T23:11:00.003

3-l is not necessary. It's implied by -a – Guido – 2014-05-17T15:27:37.550

The following line should do the trick for that:

rsync --partial --progress --rsh=ssh -r me@host.com:/datafiles/ ./

I've never used this for recursive directories before, but when I texted it just now it seemed to work as expected.

Tyler

Posted 2010-06-01T15:52:16.867

Reputation: 211

1This works too, as long as you include the -r =) – physicsmichael – 2010-06-01T17:07:17.873

Most of my rsync transfers are of the type "rsync -ahP something/ user@host:/somewhere/". The --rsh=ssh is completely unnecessary. Then, also -P drags in --partial and --progress. – killermist – 2013-06-05T01:56:03.423

The rsync command you listed would work, if you only added "-r". but you would also most likely want "-a" and "-v".

And about the trailing slash, me@site.com:/data/ is equivalent to /data/*, in other words, if you add a slash, it copies all the contents. but me@site.com:/data would be the directory itself [and naturally, its subfolders]

Matt

Posted 2010-06-01T15:52:16.867

Reputation: 627

Merging kibitzer's and Tyler's answers worked best for me:

rsync --partial --progress -avzl -e ssh /directory/with/files/ you@host.com:/new/directory/

(rsyncing recursively and showing progress of each file)

Tarc

Posted 2010-06-01T15:52:16.867

Reputation: 161