flux-fsck(1)

SYNOPSIS

flux fsck [OPTIONS]

DESCRIPTION

The flux fsck checks the integrity of the KVS backing store. By default, it starts with the most recent checkpoint (root version) written to the backing store.

OPTIONS

-h, --help: Summarize available options.

-v, --verbose: List keys on stderr as they are validated.

-q, --quiet: Reduce output to essential information. Can significantly reduce output if output will be logged.

-r, --rootref=BLOBREF: Normally the check starts with the blobref in the most recent KVS checkpoint. This option directs flux-fsck to start at an arbitrary point. BLOBREF must refer to an RFC 11 tree object of type "dir".

-R, --repair

Remove any dangling references found in KVS metadata. If a KVS value changes as a result of this repair, the key is moved to the lost+found directory. The original key is unlinked. If a key points to a KVS directory that is no longer valid, it is unlinked. All damaged keys and their disposition are listed on stderr. This process creates a new root reference for these changes, and commits it as the current KVS checkpoint at the end of the scan. The KVS is required to be unloaded during repair.

This option should be considered EXPERIMENTAL at this time.

-j, --job-aware

When specified with --repair, if a single key within a job's directory is corrupted, all data from the job will be moved to the lost+found directory.

This option should be considered EXPERIMENTAL at this time.