FSImage Loading Takes More Time

Issue

FSImage processing in Pulse takes more time (for example, more than 10 hours) to complete for large FSImages (for example, ~40 GB).


Resolution

Update FS Analytics and Elasticsearch configuration

Apply the following configuration changes to improve the performance of FSImage parsing and Elasticsearch ingestion.

  1. Step 1: Verify configuration files

On the Pulse node, check whether the following files exist at:

/data01/acceldata/config/docker/addons
  • ad-fsanalyticsv2-connector.yml

  • ad-fs-elastic.yml

  1. Step 2: Generate configuration files (if they do not already exist)

If the files do not exist, generate them using the following commands:

accelo admin makeconfig ad-fsanalyticsv2-connector accelo admin makeconfig ad-fs-elastic
  1. Update environment variables

Open the configuration files and locate the environment section in each file.

  • FS Analytics (ad-fsanalyticsv2-connector.yml)

Add the following entries:

FSANALYTICS_ENABLE_OFFHEAP_MAP=true FSANALYTICS_IMAGE_PARSING_BATCHSIZE=400000 ES_REFRESH_INTERVAL=1s
  • Elasticsearch (ad-fs-elastic.yml)

Add the following entry:

thread_pool.write.queue_size=10000

Result

These settings significantly improve FSImage parsing speed and Elasticsearch ingestion throughput, reducing overall FSImage processing time.