Metadata snapshot does not indicate an error even when disk is full #920

New issue

Open

opened 2025-01-04 22:18:01 +00:00 by baptiste · 0 comments

baptiste commented

2025-01-04 22:18:01 +00:00

Owner

I ran a metadata snapshot on all nodes of a Garage cluster, and it said "OK" for each node:

# garage meta snapshot --all
6ea290fbe1cbf9d9  ok
731a8fb9c7881547  ok
8c8a4ab1878f5f80  ok

However, one of the node ran out of disk space during the snapshot. As a result, Garage generated a snapshot file of only 700 MB while the original LMDB database is 3.4 GB.

I'm not sure how such an error should be handled (deleted the incomplete snapshot?), but at least the error should be reported.

I ran a metadata snapshot on all nodes of a Garage cluster, and it said "OK" for each node: ``` # garage meta snapshot --all 6ea290fbe1cbf9d9 ok 731a8fb9c7881547 ok 8c8a4ab1878f5f80 ok ``` However, one of the node ran out of disk space during the snapshot. As a result, Garage generated a snapshot file of only 700 MB while the original LMDB database is 3.4 GB. I'm not sure how such an error should be handled (deleted the incomplete snapshot?), but at least the error should be reported.