Skip to content

Conversation

@hvlad
Copy link
Member

@hvlad hvlad commented Nov 30, 2025

No description provided.

@hvlad hvlad self-assigned this Nov 30, 2025
@hvlad hvlad linked an issue Nov 30, 2025 that may be closed by this pull request
@dyemanov dyemanov self-requested a review December 1, 2025 09:42
@hvlad hvlad merged commit c642236 into v4.0-release Dec 2, 2025
12 of 16 checks passed
hvlad added a commit that referenced this pull request Dec 3, 2025
This should fix bug #8817 : Fatal lock manager error: invalid lock id
@hvlad hvlad deleted the work/gh-8817 branch December 3, 2025 12:20
hvlad added a commit that referenced this pull request Dec 3, 2025
This should fix bug #8817 : Fatal lock manager error: invalid lock id
@dyemanov
Copy link
Member

Unfortunately, this commit may cause a hang inside MET_scan_relation. If the new server version adds some virtual table and we open a database from a prior version (that does not contain this table), the new virtual table exists in the att_relations array and MET_scan_relation is called for it, but since it's missing inside RDB$RELATIONS, the REL_scanned flag is never set and we have an infinite loop.

@hvlad
Copy link
Member Author

hvlad commented Dec 10, 2025

Looks like REL_scanned flag should be set for system relation even if it was not found in RDB$RELATIONS ?
And raise isc_relnotdef for non-system relation ?

@dyemanov
Copy link
Member

Looks like REL_scanned flag should be set for system relation even if it was not found in RDB$RELATIONS ? And raise isc_relnotdef for non-system relation ?

Sounds like a plan.

@hvlad
Copy link
Member Author

hvlad commented Dec 15, 2025

Patch is at #8826

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Fatal lock manager error: invalid lock id

2 participants