Mandi! Martin Simmons
In chel di` si favelave...
>> OK. Because data get copied/linked BY .sync dir, option 'i' can be kept,
>> because inode does notchange.
> Yes.
OK, now seems to work. But incremental took a very huge amount of time.
Currently i have on job:
Accurate = yes
and on fileset:
# Inclusione relativa al sync spool di rsnapshot.
Include {
Options {
Signature = MD5
accurate = ipugsmd5
}
File = "/rpool-backup/rsnapshot/.sync"
}
and currently i have for incremental:
31-Jan 18:21 svpve3-sd JobId 16548: Elapsed time=19:13:51, Transfer rate=1.895
M Bytes/second
FD Files Written: 48,539
FD Bytes Written: 131,229,041,978 (131.2 GB)
while a full:
28-Jan 14:45 svpve3-sd JobId 16386: Elapsed time=41:46:29, Transfer rate=41.73
M Bytes/second
FD Files Written: 5,944,340
FD Bytes Written: 6,275,416,885,923 (6.275 TB)
So it is clear that an incremental backup have to check a roughly six
millions of files every time, but... there's some way to 'tradeoff' accuracy
versus speed tackling with 'accurate = ipugsmd5'?
EG, if i modify/reduce the number of checks, this can provide more speed?
Thanks.
--
C'era un tempo in cui l'Italia era sull'orlo di un precipizio.
Ma grazie a Berlusconi abbiamo fatto un passo avanti. (?)
_______________________________________________
Bacula-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/bacula-users
