summaryrefslogtreecommitdiff
path: root/doc/todo/speed_up_transfers_over_ssh+rsync_--_directly_reuse_the_same_connection__63__/comment_5_fca151591d0c6c75013711cb5de81d47._comment
blob: a7130ea749ae6d3f615aa3a5dc1d4e0bc588d352 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
[[!comment format=mdwn
 username="joey"
 subject="""comment 5"""
 date="2015-08-11T17:27:23Z"
 content="""
AFAIK, it's not possible to run multiple separate rsync transfers over a single
connection; rsync closes the connection after a transfer is complete (and
the rsync protocol is nowhere documented, and has only one implementation
so there's no way to avoid this behavior). 

So, rsync would need to be told a whole list of files to transfer in one go,
which poses many difficulties, including for git-annex's progress display,
and for needing to queue up a set of files when eg `git annex get` is picking
which ones to get.

I'd want to see a lot more measurements of where bottlenecks are in
the current system, before seriously considering such a thing.
"""]]