Maintenance details

Hello,

In Avamar 7, There are two time frames, Backup window and Maintenance window. During the maintenance window, a limit of eight backups per storage node is enforced. However, active backup will not be canceled if they are more than eight.

Garbage Collection is a process of removing unused chunks from backup that have been expired. This frees up capacity.

It always starts daily at beginning of Maintenance window. Backup can run during this process as per limit mentioned.

Checkpoints are read only snapshots of the avamar server. The avamar server internally manages checkpoint creation. There are two checkpoints created daily, first during maintenance window and second towards the end of maintenance window.

HFC check is a process to validate the checkpoint. There are two ways it does HFC check -

1. Full

2. Rolling

If HFC check detects any error, it automatically try to repair. By default, a rolling HFC check is automatically runs once a day.

Hope this will help you.

Regards,

Pawan

Hello,

Earlier versions of Avamar (Before Avamar 7) has Blackout window. GC runs during this time. Rest of things are similar to Avamar 7.

To understand total number of backups during Backup window, Blackout window and Maintenance window for Avamar 6.1, please refer to -

How many simultaneous client sessions can be made to the Avamar server

For Avamar 7, you can run upto 72 backups per node during backup window.

For all nodes, you can run one restore. For example if you have three nodes, you can run total 3x72 - 1 = 215 backups.

During Maintenance windows, you can run 8 backups each node as I mentioned in my previous post. Other backups will be in waiting state.

Regards,

Pawan

Yes very helpful answer.

Can you give me same information in Avamar 6.

And the other question is "How many backups & restores can we run per a node in Backup windows and maintenance window", Please specify number for Avamar 6 and And avamar 7

Hello Pawan,

Just for clarification. When the maintenance windows starts,

          1) GC will start

          2) Next Check Point

          3) HFS

          4) CheckPoint.

I would like to know what happens exactly at the first check point and second point. And also if any system corrupted to which checkpoint we can recall?

Please let me know if my question is not clear.

Hi,

During the maintenance window, the Avamar server runs GC, first Checkpoint, Checkpoint Validation (HFS Check) and a final checkpoint.

HFS check validates only first checkpoint. In case of any issue, you should roll back to validated checkpoint.

If you have any maintenance or other critical activity to be performed on Avamar server, always make sure you have validated checkpoint available. You can use cplist command to see list of all checkpoints. The checkpoint list appears similar to the following example:

cp.20140106170113 Fri Jan 6 17:01:13 2014 valid hfs del nodes 4 stripes 396 cp.20140107170042 Sat Jan 7 17:00:42 2014 valid hfs del nodes 4 stripes 396 cp.20140108170040 Sun Jan 8 17:00:40 2014 valid hfs ... nodes 4 stripes 396 cp.20140109170043 Mon Jan 9 17:00:43 2014 valid hfs ... nodes 4 stripes 396

where:

1. cp.yyyymmddhhmmss is the checkpoint ID.

2. valid hfs indicates a validated checkpoint.

If there is no validated checkpoint available, you can validate a checkpoint using command line or you can create a new checkpoint and validate. From Avamar 7.1, checkpoint validation option has been removed from GUI.  You should do it from command line. (It is recommended to contact EMC support to validate a checkpoint.)

You can delete checkpoints to reclaim additional server storage capacity. Generally, it is best to delete unvalidated checkpoints before you delete validated checkpoints.

I hope that will be helpful.

Regards,

Pawan

The statement "To rollback, the first option should always be the latest validated checkpoint."

Is incorrect.

You should always attempt to roll back to the most recent checkpoint, validated or not, otherwise you will most likely be loosing more data than is necessary.

Most checkpoints are valid regardless of if they are "Validated" or not. Start with the most recent CP and work backwards in time to the next most recent if the Rollback is unsuccessful.

I agree and disagree at the same time. Isn't the decision to pick the  right CP more dictated by what situation we are in ?

You are absolutely correct, it really depends on the situation.

For example, if you are experiencing HFSCheck issues or stripes that won't come online, then The Recommendation would be to roll back to the latest Validated CP. But if the GSAN just failed to Shutdown cleanly or start up, in most cases the most recent CP will be just fine.

Whenever you rollback to a CP that has not been validated, then you should perform another HFSCheck to make sure all is OK.

However these two cases aren't always true. That is why it is best to get Remote Reactive Customer Support on the line to determine the best course of action, depending on the circumstances.

Thank you very much to all.

I got the information which is more then required.

Good to know that, please mark helpful/correct answer, to all responses which really have been helpful to you

Regards,

Pawan

Some added Clarification in reading the check point status:

The Field that says "valid" indicates that the Checkpoint process completed properly, indicating that it is a "Valid" checkpoint, NOT "Validated".

The second field that says "hfs" is the HFSCheck field, where "hfs" indicates a full hfscheck completed successfully and the checkpoint is "Validated". If the second field says "rol", it indicates that a rolling  HFSCheck has completed successfully and the check point is "Validated".

If the second field has dashes "---", this indicates that the HFSCheck process has not run on this checkpoint and the the checkpoint  is not "validated" even though the status says "valid ---".

The second HFS field can also have some other, seldom seen, status values, which are documented in the Administration Guide. But if you should see something else there, other than what is described here, it would be advisable to contact Remote Reactive Avamar Customer Support to investigate.