Hi Rsyncers,
I have now succesfully deployed rsync as a means to replicate our main
production servers running SCOMBAG^D^D^D^D OpenServer. (Dont worry, we'll
let the company die, and switch to something better..)
To do so however, I had to split the dirs I sync (25GB orso), into smaller
chuncks so it wont barf out regulary. I use filtering for that, and it
works pretty well. At least, that was what I thought, as large jobs would
barf out more often, than small jobs.
It would stop, with an unexpected disconnect:
2003/11/14 20:21:46 [29414] rsync error: error in rsync protocol data
stream (code 12) at io.c(165)
when using --verbose --verbose --stats
In the process I noticed a lot less barf outs, when not using --stats.
It occured to me --stats might cause this, because of some negative
results in the feedback, which in itself are weird:
wrote 1747298 bytes read 116177995 bytes 81864.14 bytes/sec
total size is -627215474 speedup is -5.32
So, why the negatives? Could it cause --stats to div by zero or something?
Unfortunatly, after a earlier suggestion to trace the process, I managed
to Panic the kernel in a split second, when running "trace -o file.out
rsync...." so no more info from that angle.
Thanks for any insights,
Richard
