WARN CommitIterator: missing object

[sakalouski]

Expected Behavior

I have been accessing siva files and get WARN messages. The question is whether it happens due to corrupted data or tools? Are these warnings expected to be reproducible for the same repo? Because for now I get different number of entries for each run.

Current Behavior

WARN CommitIterator: missing object
and another one is
Missing Commit

Your Environment (for bugs)

• Spark version: 2.2.1
• PySpark version (if using PySpark): 2.2.1
• jgit-spark-connector version: 1.0.0
• Operating System and version: CentOS 7
• Some needed resources to reproduce the problem:

[bzz]

@sakalouski sorry for dropping this for so long 😱

is it happens due to corrupted data or tools

Usually, in my experience this is the case of corrupted data.

Are these warnings expected to be reproducible for the same repo

I belive they are. Is there a chance you could share the repo that triggers it?

What I did before to validate this, is to extract the .pack files from the .siva and use git plumbing commands like

siva unpack 9279be3cf07fb3cca4fc964b27acea57e0af461b.siva .
git verify-pack -v objects/pack/<>.pack
git cat-file -p <sha>