Skip to main content
Solved

Offload jobs to Service Provider that freeze Veeam Database- PostgreSQL

  • December 19, 2025
  • 7 comments
  • 83 views
Jean.peres.bkp
Forum|alt.badge.img+8

The database simply freezes and Veeam needs to be restarted.

 

We discovered that disabling Offload for the current Service Provider completely stops the problem from occurring.

 

  • VBR RAM/CPU resources remain without overload.
  • The database loses communication, and as a solution, we need to restart the postgres service.
  • Veeam Version 12.3.2.4165
  • PostgreSQL Version 15.14-1

 

We have implemented these changes so far:

 

Set-VBRPSQLDatabaseServerLimits

https://helpcenter.veeam.com/archive/backup/120/powershell/set-vbrpsqldatabaseserverlimits.html

 

AV & Firewall

Running & Installing PostgreSQL On Native Windows - PostgreSQL wiki

 

We discovered that disabling Offload for the current Service Provider completely stops the problem from occurring.

 

Has anyone else experienced a similar problem?

 

I still don't understand why Offload locks the database.

I'm looking at the database level, searching for Buffer Maxwritten Clean alerts and the like.

 

  • The workloads sent to the Service Provider are Always-On and VMware VMS.
  • Nothing too special or outdated; the whole environment is up-to-date.

 

I'm also trying to understand the availability of parallelism on the service provider's side, perhaps due to a lack of active threads and job enqueues.

 

So, if anyone has experienced this, please let me know.

Thank you!

 

Best answer by Jean.peres.bkp

To conclude this case. Summary follows:

PostgreSQL:
- Reduced max_connections from 3000 to 200.
- Adjusted timeout for inactive sessions exceeding 1 hour.
- Reduced total memory allocation to avoid exhausting all server memory.
- Configured bg writer and autovaccum.

 

O.S:

https://learn.microsoft.com/en-us/troubleshoot/windows-server/performance/slow-performance-when-using-power-plan

The power plan needs to be set to High Performance.

 

The databases freezing again.  

 

We just change the Immediate Mode to Periodic Mode Copy Job.

We've been running with these changes for a week now without any issues.

Jean Sczepanski Peres

    7 comments

    Marcel.K
    Forum|alt.badge.img+11
    • Marcel.K
    • Veeam Legend
    • December 19, 2025

    Hi ​@Jean.peres.bkp,

    I have same problem. We have been migrated DB from MS to Pqsl and we are loosing connection to the database.

    I have applied same. Limits and AV as you mentioned above.

    Console has delay and veeam is loosing connection. Reboot is needed 2-3 times per week.  

    We have same isue with copy jobs. Once they are disabled. Is running, but still not like on MS-SQL.

    Log of PSQL are ok.

    So, yes, looks thst tunning of psql config is needed ... but did not find rootcause.

    Marcel Kačmár | VCSP Partner 2022 | VMCE 2024 | VMCA 2024 | Veeam Legend 2025 |
      Jean.peres.bkp
      Forum|alt.badge.img+8
      • Jean.peres.bkp
      • Author
      • Veeam Legend
      • December 22, 2025

      ​@Marcel.K , Thank you for your contribution.
      It's interesting to find an environment with exactly the same behavior.

       

      We just follow this steps:

      PostgreSQL:
      - Reduced max_connections from 3000 to 200.
      - Adjusted timeout for inactive sessions exceeding 1 hour.
      - Reduced total memory allocation to avoid exhausting all server memory.
      - Configured bg writer and autovaccum.

       

      O.S:

      https://learn.microsoft.com/en-us/troubleshoot/windows-server/performance/slow-performance-when-using-power-plan

      The power plan needs to be set to High Performance.

       

      We'll keep you updated if things go back to normal.

      See you soon.

      Jean Sczepanski Peres
        Marcel.K
        Forum|alt.badge.img+11
        • Marcel.K
        • Veeam Legend
        • December 22, 2025

        Well, I can provide more information in new year to try to "play" with settings as I am off ...

        I will be glad if you can share your results and i can share then mine ...

        Marcel Kačmár | VCSP Partner 2022 | VMCE 2024 | VMCA 2024 | Veeam Legend 2025 |
          Jean.peres.bkp
          Forum|alt.badge.img+8
          • Jean.peres.bkp
          • Author
          • Veeam Legend
          • January 13, 2026

          Dear ​@Marcel.K  , how are you?  

           

          The databases freezing again.  

          We just change the Immediate Mode to Periodic Mode Copy Job.

          We haven't had any more outages since the 9th.

          I'll keep you updated in the coming weeks as well.

          Jean Sczepanski Peres
          Marcel.K
          Forum|alt.badge.img+11
          • Marcel.K
          • Veeam Legend
          • January 15, 2026

          Hi ​@Jean.peres.bkp,

          congtras :)

           

          we have been overprovisioned VBR servers - from 8vcpu to 16-32 and from 32-64GB to 128GB

          work memory set to 128MB, 3000 connections, shared buffer 32GB, effective cache 64-80GB

          hostssl enabled in hba conf for connection with veeam services - this was not done by via veeam dump

          now is less issues

          we see that database connection is stuck (because port is overloaded and is not answering) jobs will fail and in 1-2 minutes is back and retry jobs will continue with backups - so no impact on customer anymore

          we see that for VBRs with bigger amount of VMs (2500 VMs+) only

          i am very interested from your finding ...

           

           

          Marcel Kačmár | VCSP Partner 2022 | VMCE 2024 | VMCA 2024 | Veeam Legend 2025 |
            Jean.peres.bkp
            Forum|alt.badge.img+8
            • Jean.peres.bkp
            • Author
            • Veeam Legend
            • January 16, 2026

            Great news, ​@Marcel.K 

            We've been running with these changes for a week now without any issues.

            Jean Sczepanski Peres
            Jean.peres.bkp
            Forum|alt.badge.img+8
            • Jean.peres.bkp
            • Author
            • Veeam Legend
            • Answer
            • January 16, 2026

            To conclude this case. Summary follows:

            PostgreSQL:
            - Reduced max_connections from 3000 to 200.
            - Adjusted timeout for inactive sessions exceeding 1 hour.
            - Reduced total memory allocation to avoid exhausting all server memory.
            - Configured bg writer and autovaccum.

             

            O.S:

            https://learn.microsoft.com/en-us/troubleshoot/windows-server/performance/slow-performance-when-using-power-plan

            The power plan needs to be set to High Performance.

             

            The databases freezing again.  

             

            We just change the Immediate Mode to Periodic Mode Copy Job.

            We've been running with these changes for a week now without any issues.

            Jean Sczepanski Peres
            Terms & ConditionsAccessibility statement