cancel
Showing results for 
Search instead for 
Did you mean: 
Highlighted

Understanding schedules by frequency

After reading the documentation that I have found about frequency programming I still do not understand how they work.

Simple example that I don't understand

In a policy I have three schedules. All type Full.

  • Full (every day) - Frecuency 1 day (tested too with 22h)
  • Full (every two days) - Frecuency 2 days (tested too with 46h)
  • Full (every four days) - Frecuency 4 days (tested too with 92h)

Expected result?

  1. day1 - Full (every four days)
  2. day2 - Full (every two days)
  3. day3 - Full (every day)
  4. day4 - Full (every two days)
  5. day5 - Full (every four days)
  6. day6 - Full (every two days)
  7. day7 - Full (every day)

Obtained result

  1. day1 - Full (every four days)
  2. day2 - Full (every day) - Why? "Full (every two days)" has not run and must be preferential???
  3. day3 - Full (every two days)
  4. day4 - Full (every day)
  5. day5 - Full (every four days)
  6. day6 - Full (every day) - Why? "Full (every two days)" must be preferential???
  7. day7 - Full (every two days)

Any help is welcome

 

Tags (1)
6 Replies
Highlighted

Re: Understanding schedules by frequency

Sorry, I don't understand either. 

Why multiple FULL schedules in the same policy?

Can you list the policy config (bppllist <policy-name> -U ) and post the 'schedule' section of the output? 

Highlighted

Re: Understanding schedules by frequency

This is the schedule section of Sample Policy

 Schedule:              Full-1dia
    Type:                Full Backup
    Frequency:           every 1 day
     Excluded Dates----------
        No specific exclude dates entered
        No exclude days of week entered
    Synthetic:           0
    Checksum Change Detection: 0
    PFI Recovery:        0
    Maximum MPX:         1
    Retention Level:     0 (1 week)
    Number Copies:       1
    Fail on Error:       0
    Residence:           (specific storage unit not required)
    Volume Pool:         (same as policy volume pool)
    Server Group:        (same as specified for policy)
    Residence is Storage Lifecycle Policy:         0
    Daily Windows:
          Sunday     11:10:00  -->  Sunday     14:10:00
          Monday     11:10:00  -->  Monday     14:10:00
          Tuesday    11:10:00  -->  Tuesday    14:10:00
          Wednesday  11:10:00  -->  Wednesday  14:10:00
          Thursday   11:10:00  -->  Thursday   14:10:00
          Friday     11:10:00  -->  Friday     14:10:00
          Saturday   11:10:00  -->  Saturday   14:10:00

  Schedule:              Full-2dias
    Type:                Full Backup
    Frequency:           every 2 days
     Excluded Dates----------
        No specific exclude dates entered
        No exclude days of week entered
    Synthetic:           0
    Checksum Change Detection: 0
    PFI Recovery:        0
    Maximum MPX:         1
    Retention Level:     0 (1 week)
    Number Copies:       1
    Fail on Error:       0
    Residence:           (specific storage unit not required)
    Volume Pool:         (same as policy volume pool)
    Server Group:        (same as specified for policy)
    Residence is Storage Lifecycle Policy:         0
    Daily Windows:
          Sunday     11:10:00  -->  Sunday     14:10:00
          Monday     11:10:00  -->  Monday     14:10:00
          Tuesday    11:10:00  -->  Tuesday    14:10:00
          Wednesday  11:10:00  -->  Wednesday  14:10:00
          Thursday   11:10:00  -->  Thursday   14:10:00
          Friday     11:10:00  -->  Friday     14:10:00
          Saturday   11:10:00  -->  Saturday   14:10:00

  Schedule:              Full-4dias
    Type:                Full Backup
    Frequency:           every 4 days
     Excluded Dates----------
        No specific exclude dates entered
        No exclude days of week entered
    Synthetic:           0
    Checksum Change Detection: 0
    PFI Recovery:        0
    Maximum MPX:         1
    Retention Level:     0 (1 week)
    Number Copies:       1
    Fail on Error:       0
    Residence:           (specific storage unit not required)
    Volume Pool:         (same as policy volume pool)
    Server Group:        (same as specified for policy)
    Residence is Storage Lifecycle Policy:         0
    Daily Windows:
          Sunday     11:10:00  -->  Sunday     14:10:00
          Monday     11:10:00  -->  Monday     14:10:00
          Tuesday    11:10:00  -->  Tuesday    14:10:00
          Wednesday  11:10:00  -->  Wednesday  14:10:00
          Thursday   11:10:00  -->  Thursday   14:10:00
          Friday     11:10:00  -->  Friday     14:10:00
          Saturday   11:10:00  -->  Saturday   14:10:00


The questión is based in an example to try to undestand use of Frecuency, our current real need is to make the following Full backups

  • 1 Full Backup weekly - SLP Retention 6 months
  • 1 Full Backup monthly - SLP Retention 3 years (When a monthly Backup is executed, the weekly should not be executed since the monthly has higher retention)
  • 1 Full Backup every 3 months with the intention of running force Accelerator rescan (every 3 months) - SLP Retention 3 years (When a 3 months Backup is executed, the weekly and monthly should not be executed)

Regards.

Javier

 

Re: Understanding schedules by frequency

@JavierR 

Please have a look at my article about Frequency based scheduling:

https://vox.veritas.com/t5/Articles/NetBackup-Frequency-Based-Scheduling/ta-p/810193

There used to be an excellent flow-diagram in NBU 5.1 Admin Guide II that helped me understand how the scheduler works. For some reason this section was totally dropped from subsequent manuals.

I have posted a screenshot of the diagram in above article. 

Have a good look at the NOTES section: 

If more than one schedule for this policy is due for a client, the backups from the schedule that is backed up least often are added first.

About having Weekly and Monthly FULL backups in the same schedules:

Veritas recommends not to add Frequency and Calendar schedules in the same policy, as result may be unpredictable.

I have done this and many other users as well. This is how we prevent 2 x (or more) FULL backups from running at the same time. 

Example :
Weekly FULL backups start every Friday at 18:00.  (Frequency schedule)
Monthly FULL backups must run every LAST Friday of the month, also starting at 18:00. (Calendar schedule)
So, to ensure that only the Monthly schedule will run on the last Friday of the month, change the backup window of the Monthly Full to open 10 minutes earlier (17:50). 

The Monthly FULL will start before 18:00. 
When the Weekly FULL is due, the scheduler (nbpem) will not run the Weekly backup, because another FULL is already running. 

I hope this helps....

Highlighted

Re: Understanding schedules by frequency

Hi Marianne. 

Thank you very much for your help.

All the indicated schedules are by Frequency. There are none per Calendar. Both in the real case and in the example of the first post are schedules by Frequency.

According to the note of the referenced article: "If more than one schedule for this policy is due for a client. the backups from the schedule that is backed up least often are added first." There are no schedules with the same frequency value. Seems easy.

Three Total Backups by frecuency i same windows: Saturday 15:00 - 23:00

  • Full_Acc_3Months - Frecuency: 87 days (with force rescan accelerator) - SLP1
  • Full_monthly - Frecuency: 27 days  - SLP1
  • Full_weekly - Frecuency: 5 days - SLP2

Obtained result:

  • First backup: Full_Acc_3Months
  • Rest of weekly Backups: Full_weekly
  • Backup never executed after 6 weeks: Full_monthly

Since we do not have months to wait for the tests, we simplify with the tested Daily Frequency programming, originally exposed in the first question (with results). Same Windows for all Schedules: all days at same time. Same SLP.

  • Full (every day) - Frecuency 1 day (tested too with 22h) 
  • Full (every two days) - Frecuency 2 days (tested too with 46h) 
  • Full (every four days) - Frecuency 4 days (tested too with 92h) 

The result does not seem to conform to the rules. Some rule I'm probably not understanding.

Highlighted

Re: Understanding schedules by frequency

Looking at your test schedules and real-life experience, I agree with you - if all schedules are new and nothing has run previously, then the 'every-4-days' is expected to run first. ( backups from the schedule that is backed up least often are added first.)
On day 2, I would expect every-2-days to run next. 
every-4-days is ignored, because backup ran within last 4 days.
Between every-2-days and every-1-day, the choice should've been 'the least often'. 

If this was my choice, I would kick-start the process manually and phase in the additional schedules. 
Once each schedule ran at least once, the frequencies should 'behave' as expected. 

I would start with only Weekly Full and add the Monthly and 3-Monthly shedules shortly before I want them to run.
Be online when the 1st monthly should start. Make sure the Monthly starts, not Weekly. Even if you have to manually kick off the schedule a few minutes early. 
Add the 3-Monthly schedule a day or 2 before the schedule is due.

Once each schedule ran when they were supposed to, they should 'behave' and run when supposed to. 

I do realize that in a large environment, this will be time consuming, but maybe worth the effort? 

Highlighted

Re: Understanding schedules by frequency

This issue made me want to dig in a bit more, as I expected the same results as you and Marianne. I came across This section from the Admin Guide.

I think what is adding confusion is this part:

NetBackup determines the due time for each schedule. The due time depends on the following:

  • The last backup data for each schedule based on comparable schedules.

  • The frequency that is added to each schedule to determine which schedule is due next.

From the same section:

NetBackup makes the following comparisons to identify a comparable schedule:

Full schedules

Compared to other full schedules of the same or longer frequency.

 

So it looks like what's happening is your longest frequency schedule (4 days) kicks off first, as expected. The next day, NBU checks for what schedule to run next, and is using this schedule to compare backup data. That data shows that a full backup was completed for schedule Full-4dia, and uses that backup as its first reference point, as they're all full backups. Since it had not been 46 hours since the last comparable schedule (full) ran, it goes to the Full-1dia schedule. On the third day, it sees that the Full-2dia schedule has not run and initiates that schedule.