Skip to content

Table' names without relids #46

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 47 commits into from
Closed

Table' names without relids #46

wants to merge 47 commits into from

Conversation

Alena0704
Copy link
Collaborator

No description provided.

danolivo added 30 commits June 13, 2021 14:01
…rgeAppend, a SubqueryScan nodes.

Tags: shardman.
Replace find_ok assert with elog panic message.
Improve the elog panic message introduced in previous commit.
Fix ForeignScan estimation logic.
Switch from AQO_EXPLAIN define to the two user GUCS: aqo.details
and aqo.show_hash.

New ExplainOneNode hook allows to add info into explain of each node.
AQO supports push-down of trivial JOIN and JOIN with mergejoinable clauses.
short-living planner memory context.

Some indentations fixed.
Big commit. To avoiding UPDATE/UPDATE/DELETE/INSERT conflicts we
use user locks on a pair (query_hash, fspace_hash) to guarantee that
only one backend can do INSERT/UPDATE/DELETE of a aqo table row.
Also we use dirty snapshot to see concurrently inserted/updated tuples.
Such conflict may cause deadlocks on end-of-transaction waiting.
Currently, if we detected such conflict, we refuse from we ML-base
changing.
collection.

Add some comments on auto tuning algorithm.
change the intelligent.sql test to avoid unsteady behaviour in the
intelligent mode.
Change the logic of a node hash showing.
Add regression test on forced stat collection mode.
Enable ICU and TAP tests in the AQO CI
Add regression test on currently unsupported planner clauses.
…f sorting table rows use md5(string) instead of string itself.

Tags: aqo.
… path.parent->subroot as the planning context for interpretation of the subpath.

Tags: aqo.
…ning mode without using AQO. In this mode we aren't going to use any such data during a statistics gathering at the end of execution. Also add more safe initialization of a predicted cardinality field. Tags: aqo.
…and remove comment. Induced by a test passing problem on the powerpc node of the buildfarm.

Tags: aqo.
…ort observation of a contrib-check in browser (By A.Lakhin report).

Tags: aqo.
danolivo and others added 17 commits June 18, 2021 09:12
test for successful passing.

Check on passing of this rule is added into CI/CD script.

Further we need to implement execution of the regression tests in
different modes of AQO repeatedly on the same instance and the same database.
parallel operation, as either the master or a worker.
Now AQO could clean global data, such of query_text at the end
of transaction. It is needed because errors during query execution
phases. We need to be sure that all such data cleaned before the next
query execution.
Interface of the add_query_text routine changed.
the query must be read-only (by the flag XactReadOnly).
Add ignore option in pg_regress schedule for non-passed tests.
It is needed to exclude problems with recursive execution induced
by cache invalidation.
established by previous execution of this plan in another query.
@danolivo
Copy link
Collaborator

Реквест содержит три десятка коммитов ( т.ч. тех, что уже есть в репозитории), конфликтует с репой и некоторые коммиты имею одинаковые имена.
Предлагаю сначала обновить форк в соответствии с апстримом, сделать ребейз, засквошить некоторые коммиты для получения внятной истории коммитов.

@danolivo danolivo closed this Aug 23, 2021
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.

2 participants