Fix gp_toolkit.__gp_aocsseg_history crash on non-aocs tables. #888
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
We try to use macro to fetch relation'name after closing the relation in error message, which causes a crash. We do have heap_close in later lines, don't need to close here. Authored-by: Zhang Mingli [email protected]
GUC gp_max_partition_level sets the maximum number of levels allowed when creating a partitioned table use the Greenplum classic syntax. Use 0 for no limit. This was removed during 12 merge. Reintroduce it to main branch. This only applies to Greenplum classic syntax, Greenplum modern syntax is not affected.
Co-authored-by: Denis <[email protected]>
GPDB_12_MERGE_FIXME in ic_tcp.c: should use WaitEventSetWait() instead of select() following the routine in ic_udpifc.c The advantage of using WaitEventSetWait instead of select is that we can monitor the specific waitevent in the pg_stat_activity view. However, we still have many select() calls in current Greenplum codes, it's no need to replace all of them with WaitEventSet, Because most of them doesn't benefit much from pg_stat_activity, keep using select() is good and simple enough. Therefore, we have decided to simply remove this fixme.
```
/* GPDB_12_MERGE_FIXME: would be nicer to store this hash in the DSM segment or DSA */
shareinput_Xslice_hash = ShmemInitHash("ShareInputScan notifications",
N_SHAREINPUT_SLOTS(),
N_SHAREINPUT_SLOTS(),
&info,
HASH_ELEM | HASH_BLOBS);
```
Shmem is static shared memory which is pre-allocated and initialized during the startup of the
postmaster. The address of static shared memory is the same between different backends.
It stores frequently accessed fixed-size data. Using static shared memory is much simpler.
DSM and DSA is dynamic shared memory which is allocated and managed dynamically
during the runtime of the postgres. The address between different backends is different.
We usually use a static shared memory to store dsm_handle and other infos to help
backends to find its address. It is suitable for temporary data storage.
shareinput_Xslice_state is small and we access the field frequently,
Instead of using the static memory to store shared_scan_id, dsm_handle and other infos,
store the shareinput_Xslice_state directly in static memory is more convenient.
So remove the fixme.
pg_base64_enc_len() and its clones overestimated the output length by up to 2 bytes, as a result of sloppy thinking about where to divide. No callers require a precise estimate, so this has no consequences worse than palloc'ing a byte or two more than necessary. We might as well get it right though. This bug is very ancient, dating to commit 79d78bb which added encode.c. (The other instances were presumably copied from there.) Still, it doesn't quite seem worth back-patching. Oleg Tselebrovskiy Discussion: https://postgr.es/m/[email protected] (cherry picked from commit d98ed080bb31fd3d46281127871b7886288686d9) Co-authored-by: Tom Lane <[email protected]>
It is definitely supported and used by differential recovery in-tree. The NOTICE was added when WAL replication was introduced into GPDB in 737e216e9e5f. We have come a long way since then. Reviewed-by: Huansong Fu <[email protected]>
my-ship-it
approved these changes
Jan 23, 2025
Move oneoffPlan check to function `BuildCachedPlan`, travel through plan list once to get all oneoffPlan, transientPlan and dependsOnRole value. Remove GPDB_96_MERGE_FIXME in plancache.c.
In GPDB, we generate LockRows plan node only when Global Deadlock Detector is enabled and the select statement with locking clause contains only one table. In such case, we can sure that there are no motions, don't bother to check gp_segment_id here.
In function `gp_toolkit.gp_move_orphaned_files`, the following
SQL will call `pg_file_rename` to move orphaned files:
```
WITH OrphanedFiles AS (
xxxxx
xxxxx
)
SELECT
OrphanedFiles.gp_segment_id,
OrphanedFiles.oldpath,
OrphanedFiles.newpath,
pg_file_rename(OrphanedFiles.oldpath, OrphanedFiles.newpath,
NULL) AS move_success
FROM OrphanedFiles
```
However, `pg_file_rename` will only be called by coordinator, which
means, in a distributed environment, segments' files can't be moved
as expected.
By explicitly updating pg_statistic, there is the possibility that autoanalyze analyzes the table between when it is set and when we check for the expected result. In this case, the intent of the test is to ensure gpsd dumps quotes. I've added a row with single quotes and removed the explicit update of pg_statistic.
…y check GDD and other services like GPCC may always keep the connections to coordinator or segments. As discussed in PR#16903, we should skip checking the idle session for orphaned files detection to avoid that we can't get a detection result due to the ERROR `There is a client session running on one or more segment.Aborting...`.
This commit is to fix issue #17180.
With this commit, we can provide more detailed information about logerrors in pg_exttable view in GP7. This is to keep consistent with older GP version.
```
testdb=# SELECT logerrors, options FROM pg_exttable WHERE reloid = 'public.test_ext'::regclass;
logerrors | options
-----------+-----------------------------
t | {error_log_persistent=true}
(1 row)
```
…errors (#17199) This commit is to fix issue #17179. External table location is stored as connected string with delimiter '|' In pg_foreign_table.ftoptions `location_uris`. If location uri contains char '|', char '|' will be mistaken as delimiter '|' and cause errors. So in this commit, when location uri contains char '|', char '|' will be escaped by '\'. And when read location uri from pg_foreign_table, location uri will be reverse escaped.
installcheck was always a no-op for pg_upgrade until PostgreSQL 15 switched tests to TAP. Remove the fixme.
Authored-by: Zhang Mingli [email protected]
reshke
approved these changes
Jan 23, 2025
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.
Fixes #ISSUE_Number
What does this PR do?
Type of Change
Breaking Changes
Test Plan
make installcheckmake -C src/test installcheck-cbdb-parallelImpact
Performance:
User-facing changes:
Dependencies:
Checklist
Additional Context
CI Skip Instructions