Veritas-bu

[Veritas-bu] NB(DC)4.5/Solaris/Calendar scheduling strangeness

2002-10-03 23:33:35
Subject: [Veritas-bu] NB(DC)4.5/Solaris/Calendar scheduling strangeness
From: sarcher AT connect.com DOT au (Suzi Archer)
Date: Fri, 4 Oct 2002 13:33:35 +1000
Additional info...

I just looked, and there were many backups that were overdue from the previous 
schedule of
1 attempt a week 
this is mainly because previously we scheduled monthlies and weeklies on 
wednesday, and
the weeklies didn't happen every 4th week or so.

eg:
(from tuesday morning)
Class: general_NT, Schedule: weekly_full, Host: snake, 12 days instead of 7
Class: general_NT, Schedule: weekly_full, Host: frog, 12 days instead of 7
Class: general_bastion, Schedule: weekly_full, Host: giraffe, 19 days instead 
of 7
Class: general_bastion, Schedule: weekly_full, Host: swan, 12 days instead of 7
Class: general_bastion, Schedule: weekly_full, Host: duck, 12 days instead of 7
Class: general_bastion, Schedule: weekly_full, Host: bee, 12 days instead of 7
Class: general_bastion, Schedule: weekly_full, Host: canary, 19 days instead of 
7
Class: general_bastion, Schedule: weekly_full, Host: rhino, 12 days instead of 7


At 01:18PM, Oct 04 2002, Suzi Archer wrote :
> Hi all,
> 
> I am seeing strange behaviour with calendar scheduling.
> (Note: NB master was upgraded from 3.4.1 -> 4.5 this time last week)
> 
> I have modified the existing schedules we had, so that weekly's and monthly's
> run on a calendar schedule
> weeklies -> Every Wednesday of the month (Thursday where window starts after 
> midnight)
> monthlies -> the 15th of each month
> 
> I changed these schedules on Monday, and expected all the weekly backups to 
> run on 
> wed/thurs night. Instead *nothing* backed up at all over that night.
> 
> Reading thru the archives, I noted many references to the "overnight quirks" 
> of schedules
> slipping due to the " backup completed" time being used to determine when the 
> next schedule is
> permitted to run, and can see that perhaps the weekies from last week took 
> overly long
> to complete and thus none of them scheduled. 
> 
> There were two flaw's to the theory.
> 1) one client had been trying valiently to do a weekly backup since monday. I 
> killed it
> each day before completion, to force it to get "in sync" with everything else 
> on Wednesday.
> It did not try on Wednesday.
> 2) if the weekly backup schedule wasn't being triggered, why did the daily 
> schedule not kick in
> 
> One thing I have noted with the calendar scheduler, is that when scheduling 
> by "days of week"
> the calendar display did not update with little green ticks on every 
> wednesday. (trivial I know
> but it seemed odd - Administration console on Solaris)
> 
> Does this sound like something any of you have seen before, or know a reason 
> for?
> 
> I'm a bit stumped, and I don't know what could be causing this.
> 
> Thanks in advance
> 
> -suz
> 
> Example of backup schedule:
> 
> Policy Name:       fred
> 
>   Policy Type:         Standard
>   Active:              yes
>   Effective date:      01/01/1970 10:00:00
>   Client Compress:     no
>   Follow NFS Mounts:   no
>   Cross Mount Points:  no
>   Collect TIR info:    no 
>   Block Incremental:   no
>   Mult. Data Streams:  no
>   Frozen Image:        0
>   Backup Copy:         0
>   Client Encrypt:      no
>   Policy Priority:     0
>   Max Jobs/Policy:     Unlimited
>   Disaster Recovery:   0
>   Residence:           (specific storage unit not required)
>   Volume Pool:         Bastion
>   Keyword:             (none specified)
> 
>   HW/OS/Client:  Solaris       Solaris8      fred
> 
>   Include:  /
>             /var
>             /opt
>             /home
>             /var/log
> 
>   Schedule:          monthly_archive
>     Type:            Full Backup
>     Maximum MPX:     1
>     Retention Level: 9 (infinity)
>     Number Copies:   1
>     Fail on Error:   0
>     Residence:       (specific storage unit not required)
>     Volume Pool:     (same as policy volume pool)
>     Calendar sched: Enabled
>       Allowed to retry after run day
>       Day 15 of month
>     Daily Windows:
>           Sunday     20:00:00  -->  Monday     01:30:00
>           Monday     20:00:00  -->  Tuesday    01:30:00
>           Tuesday    20:00:00  -->  Wednesday  01:30:00
>           Wednesday  20:00:00  -->  Thursday   01:30:00
>           Thursday   20:00:00  -->  Friday     01:30:00
>           Friday     20:00:00  -->  Saturday   01:30:00
>           Saturday   20:00:00  -->  Sunday     01:30:00
> 
>   Schedule:          weekly_full
>     Type:            Full Backup
>     Maximum MPX:     2
>     Retention Level: 5 (3 months)
>     Number Copies:   1
>     Fail on Error:   0
>     Residence:       (specific storage unit not required)
>     Volume Pool:     (same as policy volume pool)
>     Calendar sched: Enabled
>       Wednesday, Week 1
>       Wednesday, Week 2
>       Wednesday, Week 3
>       Wednesday, Week 4
>       Wednesday, Week 5
>     Daily Windows:
>           Sunday     20:00:00  -->  Monday     01:30:00
>           Monday     20:00:00  -->  Tuesday    01:30:00
>           Tuesday    20:00:00  -->  Wednesday  01:30:00
>           Wednesday  20:00:00  -->  Thursday   01:30:00
>           Thursday   20:00:00  -->  Friday     01:30:00
>           Friday     20:00:00  -->  Saturday   01:30:00
>           Saturday   20:00:00  -->  Sunday     01:30:00
> 
> 
> _______________________________________________
> Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
> http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu


-- 
Suzi Archer               @connect.com.au          /"\
Systems Group             Lvl 9, 114 Albert Rd     \ /  ASCII Ribbon Campaign
sarcher AT connect.com DOT au    South Melbourne           X   Against HTML Mail
(03) 9674 3815            VIC,  3205               / \

<Prev in Thread] Current Thread [Next in Thread>