Active IQ Unified Manager Discussions

retention of primary On-demand Backup in NMC

REGIS_GARRUCHET
3,568 Views

Hello,

I'm facing a strange behavior, that may be normal or not, but I don't know how to show

IHAC who had to create a On-demand Backup in NMC before a maintenance Window.

The protection policy used is DR Mirror then Backup.

The usual schedule:

Primary node: some hourly and daily snaphots

retention in the policy are :

hourly: 1 day

daily: 1 week

weekly: 2 weeks (usually never used since there is no weekly backup scheduled)

Secondary Node: 1 mirror per day

Tertiary node: some daily and weekly backups

retention in the policy are :

hourly: 1 day (never used)

daily: 1 week

weekly: 30 weeks

The customer decided to create an on-demand backup with retention weekly since it should replace the scheduled remote weekly backup.

After now nearly one month, the weekly on-demand backup is still on the tertiary node, which is OK but it's still on the primary node as well despite a policy retention of 2 weeks.

I wonder if this is normal or not:

I thought about 2 reasons why the weekly backup on the primary node would not be deleted:

- The retention applied is the one of the remote node instead the one of the primary node. Why would that be ?

- The weekly snapshots are only deleted when new weekly snapshots are created and in normal use, no weekly snapshots are scheduled by this customer.

Before opening a case, I would like to know if this is an expected behaviour and why.

Thank you

Regis

1 ACCEPTED SOLUTION

adaikkap
3,567 Views

Hi Regis,

     What is the retention count ?  The retention count is only available via CLI use the following cli to find the same. Only when both duration and count is exceed PM deletes a snapshot or backup version. Basically PM does a least aggressive deletion of snapshots.

dfpm policy node get -q <policy name-or-id>

Also below is the algorithm that PM uses to delete backup versions.

It is not obvious how Protection Manager determines when to delete a Protection Manager-created snapshot.

Here is the algorithm for deleting PM (Protection Manager) created snapshots.

Each PM created snapshot is categorized by PM as either a daily, hourly, weekly, monthly or unlimited.
For each category except unlimited, there is a minimum retention count and a retention duration.
These settings are used to determine if old or expired snapshots will be deleted.

  1. Are there more snapshots than the retention count?
  2. No: Do not delete any snapshots. Exit
    Yes: Continue to Step 2
  3. Are there any snapshots older than the retention period?
  4. No: Do not delete any snapshots. Exit.
    Yes: Create list of candidate-to-be-deleted snapshots that exceed the retention period. Go to Step 3.
  5. Loop through list of candidate-to-be-deleted snapshots. Is this snapshot busy?
  6. No: Delete it.
    Yes: Do not delete it.

This algorithm is started by the Protection Manager conformance checker and is run on all data sets that are not suspended.
The algorithm is run for all the snapshot categories except unlimited.
Unlimited snapshots do not ever get deleted by this algorithm.

Below is an example policy. There are retention counts and retention durations, per policy node, for hourly, daily, weekly and monthly.

> dfpm policy node get example_policy

Node ID: 1
Node Name: Primary data
Hourly Retention Count: 2
Hourly Retention Duration: 86400
Daily Retention Count: 2
Daily Retention Duration: 604800
Weekly Retention Count: 1
Weekly Retention Duration: 1209600
Monthly Retention Count: 0
Monthly Retention Duration: 0
Backup Script Path:
Backup Script Run As:
Failover Script Path:
Failover Script Run As:
Snapshot Schedule Id: 0
Snapshot Schedule Name:
Warning Lag Enabled: Yes
Warning Lag Threshold: 129600
Error Lag Enabled: Yes
Error Lag Threshold: 172800

Node ID: 2
Node Name: Backup
Hourly Retention Count: 0
Hourly Retention Duration: 0
Daily Retention Count: 2
Daily Retention Duration: 1209600
Weekly Retention Count: 2
Weekly Retention Duration: 4838400
Monthly Retention Count: 1
Monthly Retention Duration: 8467200

Regards

adai

View solution in original post

3 REPLIES 3

adaikkap
3,568 Views

Hi Regis,

     What is the retention count ?  The retention count is only available via CLI use the following cli to find the same. Only when both duration and count is exceed PM deletes a snapshot or backup version. Basically PM does a least aggressive deletion of snapshots.

dfpm policy node get -q <policy name-or-id>

Also below is the algorithm that PM uses to delete backup versions.

It is not obvious how Protection Manager determines when to delete a Protection Manager-created snapshot.

Here is the algorithm for deleting PM (Protection Manager) created snapshots.

Each PM created snapshot is categorized by PM as either a daily, hourly, weekly, monthly or unlimited.
For each category except unlimited, there is a minimum retention count and a retention duration.
These settings are used to determine if old or expired snapshots will be deleted.

  1. Are there more snapshots than the retention count?
  2. No: Do not delete any snapshots. Exit
    Yes: Continue to Step 2
  3. Are there any snapshots older than the retention period?
  4. No: Do not delete any snapshots. Exit.
    Yes: Create list of candidate-to-be-deleted snapshots that exceed the retention period. Go to Step 3.
  5. Loop through list of candidate-to-be-deleted snapshots. Is this snapshot busy?
  6. No: Delete it.
    Yes: Do not delete it.

This algorithm is started by the Protection Manager conformance checker and is run on all data sets that are not suspended.
The algorithm is run for all the snapshot categories except unlimited.
Unlimited snapshots do not ever get deleted by this algorithm.

Below is an example policy. There are retention counts and retention durations, per policy node, for hourly, daily, weekly and monthly.

> dfpm policy node get example_policy

Node ID: 1
Node Name: Primary data
Hourly Retention Count: 2
Hourly Retention Duration: 86400
Daily Retention Count: 2
Daily Retention Duration: 604800
Weekly Retention Count: 1
Weekly Retention Duration: 1209600
Monthly Retention Count: 0
Monthly Retention Duration: 0
Backup Script Path:
Backup Script Run As:
Failover Script Path:
Failover Script Run As:
Snapshot Schedule Id: 0
Snapshot Schedule Name:
Warning Lag Enabled: Yes
Warning Lag Threshold: 129600
Error Lag Enabled: Yes
Error Lag Threshold: 172800

Node ID: 2
Node Name: Backup
Hourly Retention Count: 0
Hourly Retention Duration: 0
Daily Retention Count: 2
Daily Retention Duration: 1209600
Weekly Retention Count: 2
Weekly Retention Duration: 4838400
Monthly Retention Count: 1
Monthly Retention Duration: 8467200

Regards

adai

REGIS_GARRUCHET
3,567 Views

Hi Adai,

Thank you for all these infos.

I asked for the customer to have the results of the command you gave me. I will post them when I have them.

In the mean time, I tried on my test platform with the default policy DR Mirror then Backup,

It says for Node 1 (Primary Data) that :

weeklyRetentionCount=1

So I guess, it means that there should be always 1 weekly snapshot available on the primary node.

Is "1" the default value for weekly snapshots on Primary Data  ?

If I understand well, (if the retention count for weekly snapshot is 1 as well on the customer site), in our case where this weekly  snapshot was a on-demand snapshot and there is no schedule for weekly snapshots, it means:

- it would stay forever unless we generate another snapshot or we change the retention count to 0 . Am I right ?

Best regards,

Regis

adaikkap
3,567 Views

Hi Regis,

So I guess, it means that there should be always 1 weekly snapshot available on the primary node.

Is "1" the default value for weekly snapshots on Primary Data  ?

Yes. Thats what my memory says.

If I understand well, (if the retention count for weekly snapshot is 1 as well on the customer site), in our case where this weekly  snapshot was a on-demand snapshot and there is no schedule for weekly snapshots, it means:

- it would stay forever unless we generate another snapshot or we change the retention count to 0 . Am I right ?

Yes. Even after you generate another weekly snapshot, or Set the count to 0. It would still remain until  14 days since it was created as thats the duration count.

Regards

adai

Public