Broken blobs might appear in your evr-glacier-storage because of disk corruption or other external factors. Sometimes these corruptions are detected by the client applications that verify the integrity between the blob’s sha hash and the blobs content.
In the following example evr-fs logged an issue when trying to fetch a single blob:
2025-01-02T10:29:20 fE Expected blob ref sha3-224-ff8d…30a1 did not match actual blob ref sha3-224-eb09…32d9 calculated from blob's data
Data corruptions like this should be fixed in two steps:
To remove the broken blobs from the evr-glacier-storage index you can either completely reindex buckets of the evr-glacier-storage. Stop the evr-glacier-storage server. Remove the evr-glacier-storage’s index.db file. Make sure you don’t remove any file with the evb extension. Then start the evr-glacier-storage server again. A missing index.db will trigger a complete reindex. Broken blobs will be omitted in the generated index.
Now the evr-glacier-storage knows which good blobs are left. A sync with the backup will now retrieve the broken blobs. Read the Backup chapter for more details on how to recover from backups.