cancel
Showing results for 
Search instead for 
Did you mean: 

How to allow a policy to run more than 24 hours yet also have a specific start time - want to avoid status 196

sclind
Moderator
Moderator
   VIP   

I have some VMware intelligent backup policies that run every other day (some odd days, some even days).  The policies have a start time of 18:00 and they close 23 hours late. 

Depending on the mix of jobs sometimes all the jobs in the policy do not complete and the remaining ones cancel with status 196.  But I dont ever want a submitted job to not run - its OK if they run beyond 23 hours.

I tried changing the window for the policy to include all 24 hours but then the policy kicks in at 00:00 - which I dont want. I want to keep my specified start time.

How can I have a specific start time and yet also avoid 196's?

Scott

1 ACCEPTED SOLUTION

Accepted Solutions

tunix2k
Level 5
Partner Accredited

There is no wonder or special trick. If your ressource limit doesnt allow to start all backups in the given backup windows you may grow this windows to end later than 06:00.

Another option is to think about your ressource limts. Are these limits to strong ? Any chance to change there something ?

As I understand you are backing up from 2 ESXi Clusters. All VMs from 1th cluster on even and all VMs from second cluster on odd days. I'am right ?

If so you should change your selection to backup only the half of your VMs and the second half from a new policy on next day. I mean half of VMs of 1th cluster and half of 2th cluster on even days and the second half on odd days.

K.R.

Martin

View solution in original post

17 REPLIES 17

RiaanBadenhorst
Moderator
Moderator
Partner    VIP    Accredited Certified

If you frequency schedule it will cross 00:00 and not start another job. That will only happen if you're using calendar schedule and the day is open.

 

Frequency 1 day

Start 18:00

End 06:00 (next day)

sclind
Moderator
Moderator
   VIP   

Riaan - When the VIP policy gets submitted it the parent job starts then it submits the hundreds of backup jobs (one per VM).  So I dont know what you mean when you say "not start another job".  Crossing 00:00 is not a problem; I just dont want the policy to start a 00:00. And I dont want jobs to ever cancel with status 196 (backup window closed).

I have to calender schedule because I want the jobs to run every other day (odd or even depending on the policy).

 

RiaanBadenhorst
Moderator
Moderator
Partner    VIP    Accredited Certified

I'm referring to the parent job. If you use calendar, and open the window across 00:00 (and you've selected both days in the calendar days) it will kick of the job again. Because, the schedule is open, and the day is open.

 

I think i recall your previous post about odd/even days.

 

Why don't you split the job into two? Have 50% start before 00:00, and the next job start after 00:00?

sclind
Moderator
Moderator
   VIP   

It's a VMware Intelligent backup policy so I dont really have the ability to split it in half.  

RiaanBadenhorst
Moderator
Moderator
Partner    VIP    Accredited Certified

I would run it this way

 

Sun xxxxxxxxxxxxxxxxxxYYYYYY

Mon YYYYYYxxxxxxxxxxxxxxxxxx

Tue xxxxxxxxxxxxxxxxxxYYYYYY

Wed YYYYYYxxxxxxxxxxxxxxxxxx

Thu xxxxxxxxxxxxxxxxxxYYYYYY

Fri YYYYYYxxxxxxxxxxxxxxxxxx

 

But then you're screwed on Saturday.I think i suggested this last time too. Your scenario is a bit tricky.

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

Please bear in mind that a backup windows is a period during which backups are allowed to START.

Backup window is not associated with the the length of time that the backup runs.

Status 196 means that the backup job is queued bot cannot go active due to some limitation.

You need to check job status while the backup is queued - what is it waiting for that is preventing job for going active?

Once we know what the limitation is, we can look at the resources and settings that may enable queued jobs to go active before the backup window closes.

sclind
Moderator
Moderator
   VIP   

The jobs are queued since I am using the Resource Limit feature to limit the number of jobs that hit a datastore and/or an ESXserver.  So them being queued is normal.

What I dont want is a job to cancel once its been submitted.  I want the whole VMware Intelligent Policy to complete.

 

sclind
Moderator
Moderator
   VIP   

I have half of my VMware Intelligent Policies run on odd days and the other half on even days so doing something by day of week would be difficult.

Genericus
Moderator
Moderator
   VIP   

Why not just run user submitted jobs, and control from the OS?

Set the schedule to allow 24x7 running, and using crontab or other scheduler, start them as you wish!

I know from unix it would not be too difficult to determine odd/even day, and then call the appropriate schedule at 1800, daily.

 

OR 

 

You might try frequency based scheduling and run every 48 hours, set the start at 1800 and allow 40 hours to complete. You would need multiple policies to do this, at least one for odd and one for even.

 

NetBackup 8.3.0.1 on Solaris 11, writing to Data Domain 9800 6.2.1.0
duplicating via SLP to LTO5 in SL8500 via ACSLS

sclind
Moderator
Moderator
   VIP   

Genericus - 2 very good ideas that I may need to go with.  But I was hoping to be able to do this within Netbackup.

Now that we do VMware backups using Intelligent Policies the concept of all backups completing in a 24 hour period gets sort of outdated.

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

What I dont want is a job to cancel once its been submitted.  I want the whole VMware Intelligent Policy to complete.

Again - backup window is a period during which backups are allowed to START. 
It does not affect or care about duration of the backup.
If the backup goes Active 1 minute before the window closes, it will not fail when the window closes.

Only queued backups will fail with 196 when the window closes.

It seems that you need to have another look at backup infrastructure and VM resources to ensure backups can start and complete within backup windows. 

WVT
Level 4
Partner Accredited Certified
What device are you writing backups to, and how many media servers do you have.

sclind
Moderator
Moderator
   VIP   

I am writing to LTO4 tape (Storagetek L700 silo). There is one media server that writes the VMWare backups.

WVT
Level 4
Partner Accredited Certified
Is that one media server engaging all the esx servers possible for the entire window of given day.?

sclind
Moderator
Moderator
   VIP   

My media server does Exchange backups during the day then the VMware backups kick in starting at 6 pm.

Some clusters get done on even days of the week; the others on odd days of the week.

Since we use Resource Limits the mix of the jobs changes so sometimes they get done in a 24 hour period and sometimes they dont.

WVT
Level 4
Partner Accredited Certified
Let me rephrase...on an even day are all the applicable esx clusters being backed up simultaneously up to the resource limit for entire backup window.?

tunix2k
Level 5
Partner Accredited

There is no wonder or special trick. If your ressource limit doesnt allow to start all backups in the given backup windows you may grow this windows to end later than 06:00.

Another option is to think about your ressource limts. Are these limits to strong ? Any chance to change there something ?

As I understand you are backing up from 2 ESXi Clusters. All VMs from 1th cluster on even and all VMs from second cluster on odd days. I'am right ?

If so you should change your selection to backup only the half of your VMs and the second half from a new policy on next day. I mean half of VMs of 1th cluster and half of 2th cluster on even days and the second half on odd days.

K.R.

Martin