*Summary*This patch against dirvish-1.2.1 adds support for the option `concurrent:', which accepts a number >=1 (defaults to 1) which specifies how many concurrent dirvish processes dirvish-runall should spawn.
*Background*Some very slow hosts would slow down our nightly run and cause us to miss backups for hosts that easily could have completed in time. If the limiting factor is the remote speed (either network or remote disk), rather than local disk speed, running concurrent processes speeds up the backup.
This patch only works on *nix systems (what else would you be running dirvish on ;-) ) that support forking (read: not Windows).
As soon as I get some time, I'd like to look at doing something similar to 1.3.1, but since this is still the version everyone uses, I made the changes here. Another thought is to try and use perl- style threads instead of forks.
I've been running this for several months without problem, but if anyone would like to test it on their setup or check my code for bugs I'd appreciate the feedback. Patch with -p1 in the source tree. I've been meaning to submit it but forgot until a recent post about multiple dirvish runs reminded me.
-- Eric Searcy OSU Open Source Lab
concurrency.patch
Description: Binary data
PGP.sig
Description: This is a digitally signed message part
_______________________________________________ Dirvish mailing list [email protected] http://www.dirvish.org/mailman/listinfo/dirvish
