[NIFI-11670] Encrypted Content Repository is very slow when FlowFiles have a non-zero Content Claim Offset

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.0.0-M1, 1.23.0
• Component/s: Core Framework
• Assignee: Mark Payne
• Reporter: Mark Payne
• Created: 08/Jun/23 19:22
• Updated: 20/Aug/23 20:14
• Resolved: 12/Jun/23 15:24

Description

To replicate, create a flow with GenerateFlowFile -> MergeContent

Configure GenerateFlowFile to generate 25 KB FlowFiles with a batch size of 1,000. It's important in order to replicate, that batch size be used.

Configure MergeContent to merge bins of 1,000 FlowFiles.

Merging the files when using unencrypted/default content repository takes milliseconds. Using Encrypted Repo it takes nearly a minute.

Issue Links

• causes: NIFI-11971 FlowFile content is corrupted across the whole NiFi instance throughout ProcessSession::write with omitting writing any byte to OutputStream
• fixes: NIFI-10235 Provenance replay fails when repository encryption is enabled
• relates to: NIFI-11679 Refactor EncryptedRepoContentAccessIT Configuration
• links to: GitHub Pull Request #7363