Michael Prokop

Interesting hints, thanks. No matter whether I'm running with `net.ipv4.tcp_sack=1` or `net.ipv4.tcp_sack=0`, this doesn't seem to change anything sadly. I got runtime panics though, full log available in http://michael-prokop.at/tmp/restic_panic.txt jftr.

@fd0 - jepp, did some tests today (using v0.7.1) and it looks much better \o/ Being the initial bug reporter from my PoV you can close this issue. Thanks for...

Well, _maybe_ I was too fast with my reply, just had the following failure with a 103GB directory (hetzner storagebox as repository target via sftp, restic v0.7.1 locally): ```` #...

Same problem again, this time after ~37 minutes: ``` scan [/srv] scanned 114020 directories, 1211537 files in 3:08 panic: Close: failed to send packet header: write |1: broken pipe325557 items...

FTR, retried with latest restic from git (v0.7.1-183-g0f25ef94) with debug enabled, last lines from debug log: ``` 2017/09/12 19:12:58 repository/master_index.go:39 repository.(*MasterIndex).Lookup 108502 id c79f83fb not found in any index 2017/09/12...

And another update of mine :) With latest restic from git and running it under `ionice -c 3 nice -n 19` it seems to be way better/stable so far. (Or...

We noticed, that `facter`, `hiera` + `puppet` and their dependencies require ~32MB of disk space on grml-full, I suppose that's worth keeping it? :thinking: Opinions?

I'd suggest to also drop `mcollective`, which requires just 1,084 kB disk space, but I think it doesn't provide any real usage on Grml live system?

@zeha thanks for fast feedback, then let's do this! :)

With the recent changes, we managed to get back from 876 MB to 831MB (grml64-full_testing, builds 4020 vs 4021).