Recover the pool catalog after losing the node
Pool the spare power of every computer into one cloud. Two apps (sys0
Brought to you by:
a-hackers-guide
Menu
▾
▴
#27 Recover the pool catalog after losing the node
open
nobody
2 days ago
2 days ago
Anonymous
No
Originally created by: A-Hackers-Guide
If the node's disk is lost, chunks survive on the machines but the node's list of snapshots (in node.journal) is gone, so nothing can be restored even with the recovery key.
Plan:
- [ ]
sys1 storage recover --key <recovery key>on a fresh node: ask every machine for its inventory, try to open each chunk as a snapshot root with keys derived for known owners, and rebuild the snapshot list - [ ] store an encrypted copy of the catalog itself in the pool, so recovery reads one chunk instead of scanning
- [ ] longer term, multiple nodes with replicated state (#15) remove the single point of failure
Until then: back up the node's data directory. See docs/STORAGE.md.
Originally posted by: A-Hackers-Guide
Mitigated in v0.5.0: with a 3-node cluster the catalog is replicated, so losing one node's disk loses nothing. Still open for single-node setups: rebuilding the catalog from machines' chunks with the recovery key.