CHUNK_ERASURECODE recovery

Support
#1

To replace the HDDs in my primary backup RAID5 enclosure, I ran a duplicacy copy to a single HDD configured with erasure encoding, successfully check -chunks, replaced the RAID5 HDDs, then copied the single HDD storage back to the RAID5. The copy operation succeeded, but with this warning:

2026-02-17 23:48:08.118 WARN CHUNK_ERASURECODE Recovering a 13837396 byte chunk from 2767480 byte shards: ****-*-

A subsequent check -chunks of the RAID5 succeeded. The storage contains about 1M chunks. I normally used the single HDD for a monthly cold backup that I keep offline.

Does the message mean that the chunk on the single HDD was recovered and re-written? If not, is there a procedure to correct the chunk on the HDD?

1 Like
#2

Check -chunks checks each chunk only once. If you want to recheck all chunks again you need to delete verified_chunks file.

I don’t think it fixes the issue in the original chunk. But since data was recovered and copy succeeded — you are fine. This is precisely a happy path of erasure encoding.

Your experience is an excellent illustration on how one cannot trust single hard drive for anything, for any amount of time, even if it “was fine just a minute ago”; data can, and sometimes will, rot immediately.

#3

I’d removed and re-created the RAID5 storage using the web UI. After restoring the data from the single HDD, check -chunks skipped the checks – the web UI did not delete repositories/localhost/all/.duplicacy/cache/<storage name>, leaving verified_chunks. Seems like a bug, yes?

I deleted verified_chunks and am repeating check -chunks.

The corrupt chunk isn’t identified. Is re-initializing the single HDD storage and running copy the only way to fix the corrupt chunk?

Exactly. The single HDD is a WD120EFGX purchased five months ago and used only once/month, so it’s very low-mileage.