Avamar: About the use of manual Garbage Collection

Before reading this article, be familiar with how Avamar garbage collection and space reclamation works.

Avamar space reclamation processes - Part 1: Garbage Collection
Avamar space reclamation processes - Part 2: Crunching
Avamar space reclamation processes - Part 3:  The Remove Checkpoint Process (RMCP)


On an Avamar system, garbage collection (GC) is intended to run as a scheduled process.

It is possible to run GC manually. This has historically been known as "aggressive garbage collection" (AGC). In reality, this is no different to regular GC and the term "manual garbage collection", should be used.


Manual garbage collection refers to situations where the GC maintenance task is run for an extended period of time without the use of the Avamar maintenance scheduler. 

This technique might be useful to help alleviate acute issues where the Avamar User capacity is very high, or if the system is in "read-only" mode. 

In these cases, garbage collection is run manually to bring down the capacity level to just below the read-only threshold so that the backup window can run and automated garbage collection can continue working as normal.

This article discusses where the use of manual garbage collection is appropriate and the conditions of its use.
 

The first thing to note is that Avamar can take care of itself by automated maintenance tasks. 

It should not be necessary to micromanage an Avamar system.  Wherever possible, scheduled garbage collection should be allowed to "take care of business" and clear up expired data from the system. 

The only time that manual garbage collection might be appropriate is when there is an "extremely high GSAN capacity" situation. his might happen where proper GC has been hindered over a long period of time. 

If an Avamar Support engineer has performed appropriate analysis, they may seek approval for manual garbage collection activity from an Avamar Subject Matter Expert. 

Use of manual garbage collect is not appropriate when addressing a "high OS capacity" situation or a "high GSAN capacity utilization" situation.  Where GSAN capacity is high, it crucial that scheduled GC runs every day, and for the required amount of time. 

Using the 'capacity.sh' script described in Avamar: How to use capacity.sh script to manage capacity on an Avamar system, ensure that garbage collection runs for two passes minimum. 

Extend the maintenance window gradually. This could be over a period of days. Monitor the number of passes achieved, and the amount of data removed, until the capacity trend turns negative. 

If the capacity trend does not turn negative, and the capacity does not lower, apply additional capacity management measures. See Avamar: Capacity Management Concepts and Training.

Support can help check that garbage collection is running optimally.

• A system tuned to balance incoming and expired data runs in steady-state. See the Avamar operational best practices guide. 
• Running garbage collection manually "masks" capacity issues by postponing the steps that must be taken to get a system into a healthy state for long-term uninterrupted use. 

• Running Garbage Collection manually competes with backup or replication activity for hardware resources. All-round performance suffers.
• Certain actions required for garbage collection run automatically ONLY at the scheduled maintenance window times, for example unloading of "Index Caches" to free up more resources.