[stats] persist schema hashes for purging #8674
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
@max-hoffman DOLT
|
|
@coffeegoddd DOLT
|
… into max/stats-load-schema-purge
zachmu
approved these changes
Dec 12, 2024
| @@ -65,7 +65,7 @@ func CreateCommitArgParser() *argparser.ArgParser { | |||
| ap.SupportsFlag(SkipEmptyFlag, "", "Only create a commit if there are staged changes. If no changes are staged, the call to commit is a no-op. Cannot be used with --allow-empty.") | |||
| ap.SupportsString(DateParam, "", "date", "Specify the date used in the commit. If not specified the current system time is used.") | |||
| ap.SupportsFlag(ForceFlag, "f", "Ignores any foreign key warnings and proceeds with the commit.") | |||
| ap.SupportsString(AuthorParam, "", "author", "Specify an explicit author using the standard A U Thor {{.LessThan}}[email protected]{{.GreaterThan}} format.") | |||
| ap.SupportsString(AuthorParam, "", "author", "Specify an explicit author using the standard author {{.LessThan}}[email protected]{{.GreaterThan}} format.") | |||
There was a problem hiding this comment.
This wasn't a typo, A U Thor is a term of art
| return false, err | ||
| } | ||
|
|
||
| var tags []ref.TagRef |
There was a problem hiding this comment.
I don't think this is too dumb but you should use a new ref type for this rather than reusing tags
| // ValidateSchemas returns false if any table schema in the session | ||
| // root is incompatible with the latest schema used to create a stored | ||
| // set of statistics. | ||
| ValidateSchemas(ctx *sql.Context, branch string) (bool, error) |
There was a problem hiding this comment.
Probably should call this SchemaChange() and invert the logic
|
@max-hoffman DOLT
|
|
@max-hoffman DOLT
|
|
@max-hoffman DOLT
|
|
@max-hoffman DOLT
|
|
@zachmu i added boilerplate for a |
nicktobey
pushed a commit
that referenced
this pull request
Dec 17, 2024
* [statspro] Save table schema hashes to detect alterations between stats loads * fix preexisting drop datebase issue * [ga-format-pr] Run go/utils/repofmt/format_repo.sh and go/Godeps/update.sh * fix testing harness and in-memory restarts after alter * tuple fb type * add new files * add tuple ident * fix file ident * idents again --------- Co-authored-by: max-hoffman <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
We previously would flush stats when an active server detected schema changes, but the same problem affects initializing a server's stats after offline table alters. The prototype here uses tags to persist schema hashes so that we detect differences between stored and present table changes.
Additional improvements to debug logging and error handling. Fixes a separate bug related to re-initializing database statistics after a drop within the same server lifetime.
Note: There are interruption points in-between detecting schema changes, deleting the schema tags, and deleting the data off disk. There is also an interruption point between writing a schema tag and writing the associated data. I've tried structuring the orderings so that the worst-case scenario is we have to do repeated stats collection work:
So an interrupted delete will continue to delete after the process is picked up. And an interrupted write will be missed and have to be recollected.