fix(schema): Rollback v15 ProveCommitAggregate param schema to match #907
Conversation
|
|
||
| -- network v15 network upgrade epoch | ||
| WHERE height > 1594679 | ||
| AND method = 'ProveCommitAggregate'; |
There was a problem hiding this comment.
This limits the update to only the messages in parsed_messages table (no other) since the upgrade epoch and for this specifically affected message type. The cost will grow the longer that v0.8.6-v0.8.7 is run before updating to v0.8.8.
There was a problem hiding this comment.
This is slick!
One concern I have: we will need to apply this migration to all messages with params containing bitfields, I think this include more than just the ProveCommitAggregate params. Perhaps there is a query that could find all these types?
There was a problem hiding this comment.
Yes, we can definitely do a manual search to see which types need adjusting and write a migration for each. 👍
| '{"rle":', params->'SectorNumbers'->>'RLE', | ||
| ',"elemcount":', params->'SectorNumbers'->>'Count', | ||
| ',"_type":"bitfield"}' | ||
| )::jsonb, |
There was a problem hiding this comment.
This creates a JSON literal which is cast to a jsonb. It interpolates the existing values into the schema with the original keys.
|
After discussion aside w @frrist we aren't sure whether we want to include a migration with the fix as it would allow uncaught invariants to continue existing when we can be certain of accuracy by deleting and re-walking over the affected epochs (which would be communicated via docs/notices). So for live node operators, the application of this fix looks something like:
This is being parked until our in-sync discussion tomorrow. Comments welcome. |
I too think we can live without a migration and fix manually. The most important thing are the dumps, which cannot be fixed by migration anyways. |
|
Can you create an issue with the |
placer14
force-pushed
the
mg/fix/v15-agg-provecommit-params
branch
from
e37da77 to
1fcc11a
Compare
|
We agreed to remove the migration so it could be applied as required by the use case along w documentation describing the bug (which will be limited to live data). Our data archive will correct this bug with an appropriate announcement of it's completion. |
fixes #904