> Date: Fri, 27 Jul 2007 09:14:57 -0400
> [...]
> But, aren't we getting away from the OP's question?
>
> I don't think Kate's really looking for advice about whether
> running an external scheduler is "recommended" or even
> acceptable, but if you DO run an external scheduler, are the
> policy priorities respected??
Summary: yes. :-)
Man, was this thread as painful to sit through day after day as it
was/is to try to catch up on? To Paul's Most Excellent point:
The Policy Priority field rules. An immediate backup--_obviously_ what
Kate meant since user backups aren't prioritized/run by NetBackup's
scheduler in the first place--is, AFAIK, identical to a manual backup or
one kicked off by the scheduler. Two scheduling functions apply:
policy priority and schedule priority (from the frequency interval).
But one test is worth a thousand expert opinions, so I did one.
TEST-pri1, TEST-pri2 and TEST-pri3 are identical classes except that
Policy Priority is 1, 2, 3 respectively. They back up to a test DSU
which allows only one concurrent job. They have full/cinc/diff
schedules, where the frequency of the full is 7 days and the others are
one day. This is how I ran them to exercise the scheduling logic, if
any (execution jobids appended for this writeup):
# cat /tmp/testpri
#!/bin/sh
date
cd /usr/openv/netbackup/bin
./bpbackup -i -p TEST-pri1 14829
./bpbackup -i -p TEST-pri1 14830
./bpbackup -i -p TEST-pri2 14831
./bpbackup -i -p TEST-pri2 14832
./bpbackup -i -p TEST-pri3 14833
./bpbackup -i -p TEST-pri3 14834
#
The first instance of TEST-pri1 (jobid 14829) ran, as the DSU was idle.
In order, the five that the scheduler handled were run in priority
order, then first-entered order, and the full schedule always won:
14833, 14834, 14831, 14832, 14830.
_______________________________________________
Veritas-bu maillist - Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
|