threads listlengths 1 2.99k |
|---|
[
{
"msg_contents": "While working on some other patches, I found myself wanting to use the\nfollowing command to vacuum the catalogs in all databases in a cluster:\n\n\tvacuumdb --all --schema pg_catalog\n\nHowever, this presently fails with the following error:\n\n\tcannot vacuum specific schema(s) in all datab... |
[
{
"msg_contents": "Hi all,\n\nWhile working on a different patch, I have noted three code paths that\ncall changeDependencyFor() but don't check that they do not return\nerrors. In all the three cases (support function, extension/schema\nand object/schema), it seems to me that only one dependency update is\nex... |
[
{
"msg_contents": "While working on the invalid parameterized join path issue [1], I\nnoticed that we can simplify the codes for checking parameterized\npartial paths in try_partial_hashjoin/mergejoin_path, with the help of\nmacro PATH_REQ_OUTER.\n\n- if (inner_path->param_info != NULL)\n- {\n- ... |
[
{
"msg_contents": "Background\n==========\nPostgreSQL has an amazing variety of routines for accessing files. Consider just the “open file” routines.\n PathNameOpenFile, OpenTemporaryFile, BasicOpenFile, open, fopen, BufFileCreateFileSet,\n BufFileOpenFileSet, AllocateFile, OpenTransientFile, FileSetCreate... |
[
{
"msg_contents": "Hi,\n\nIt's documented that a failed REINDEX can leave behind a transient\nindex, and I'm not going to speculate on all the conditions that could\nlead to this situation. However, cancelling a REINDEX CONCURRENTLY\nwill reliably leave behind the index it was building (<index\nname>_ccnew).\n... |
[
{
"msg_contents": "Hi,\r\n\r\nplan_create_index_workers[1] does not consider the amount of tuples \r\nexisting in TOAST pages when determining the number of parallel workers \r\nto use for a build. The estimation comes from estimate_rel_size[2], \r\nwhich in this case, will just take the value from rel->rd_rel-... |
[
{
"msg_contents": "Hi!\n\n(posting this to -hackers rather than to -docs since it seems a deeper\nproblem than just adjusting the docs)\n\nI recently observed a case with standby corruption after upgrading pg12 to\npg14, which was presented in the form of XX001 errors on the new cluster's\nstandby nodes. e.g.:\... |
[
{
"msg_contents": "I hope this email finds you well. I am excited to share that I have\nextended the functionality of the `pg_buffercache` extension by\nimplementing buffer invalidation capability, as requested by some\nPostgreSQL contributors for improved testing scenarios.\n\nThis marks my first time submitti... |
[
{
"msg_contents": "Hi hackers,\n\nAt Neon, we've been working on removing the file system dependency\nfrom PostgreSQL and replacing it with a distributed storage layer. For\nnow, we've seen most success in this by replacing the implementation\nof the smgr API, but it did require some core modifications like tho... |
[
{
"msg_contents": "Hi,\n\nWhen I read the documents and coding of SPI, [1]\nI found that the following the SPI_start_transaction does not support\ntransaciton_mode(ISOLATION LEVEL, READ WRITE/READ ONLY) like BEGIN \ncommand. [2]\nIs there a reason for this?\n\nI would like to be able to set transaciton_mode in ... |
[
{
"msg_contents": "I spotted this comment in walsender.c:\n\n> \t/*-------\n> \t * When reading from a historic timeline, and there is a timeline switch\n> \t * [.. long comment omitted ...]\n> \t * portion of the old segment is copied to the new file. -------\n> \t */\n\nNote the bogus dashes at the end. This... |
[
{
"msg_contents": "I noticed that pgrowlocks will use different names for shared locks\ndepending on whether the locks are intermediated by a multixact or not.\nParticularly, if a single transaction has locked a row, it may return \"For\nKey Share\" or \"For Share\" in the \"modes\" array, while if multiple\ntr... |
[
{
"msg_contents": "Hi,\n\nAs discussed somewhat at PgCon, enclosed is v1 of a patch to provide\nvariable block sizes; basically instead of BLCKSZ being a compile-time\nconstant, a single set of binaries can support all of the block sizes\nPg can support, using the value stored in pg_control as the basis.\n(Poss... |
[
{
"msg_contents": "I think this is the second decennial thread [0] for removing this GUC.\nThis topic came up at PGCon, so I thought I'd start the discussion on the\nlists.\n\nI'm personally not aware of anyone using this parameter. A couple of my\ncolleagues claimed to have used it in the aughts, but they als... |
[
{
"msg_contents": "Greetings,\n\nAttached please find a tarball (rather than a patch) for a proposed new \ncontrib extension, pg_stat_logmsg.\n\nThe basic idea is to mirror how pg_stat_statements works, except the \nlogged messages keyed by filename, lineno, and elevel are saved with a \naggregate count. The fo... |
[
{
"msg_contents": "Hi,\n\nI just noticed that the comment for PG_CACHE_LINE_SIZE still says that \"it's\ncurrently used in xlog.c\", which hasn't been true for quite some time.\n\nPFA a naive patch to make the description more generic.",
"msg_date": "Sat, 1 Jul 2023 15:49:36 +0800",
"msg_from": "Julien ... |
[
{
"msg_contents": "Hi,\n\nI think there's some sort of bug in how dd38ff28ad deals with\ncontrecords. Consider something as simple as\n\n pgbench -i -s 100\n\nand then doing pg_waldump on the WAL segments, I get this for every\nsingle one:\n\n pg_waldump: error: error in WAL record at 0/1FFFF98: missing\n ... |
[
{
"msg_contents": "Hi, there...\n\ndrop table infomask_test;\nCREATE TABLE infomask_test(acc_no integer PRIMARY KEY,amount\nnumeric,misc text);\nINSERT INTO infomask_test VALUES (1, 100.00,default), (2,\n200.00,repeat('abc',700));\n\nBEGIN;\nSELECT acc_no,ctid,xmin,xmax FROM infomask_test WHERE acc_no = 1... |
[
{
"msg_contents": "The buildfarm animal fairywren has been failing the tests for \npg_basebackup because it can't create a file with a path longer than 255 \nchars. This has just been tripped because for release 16 it's running \nTAP tests, and the branch name is part of the file path, and \n\"REL_16_STABLE\" i... |
[
{
"msg_contents": "Hi,\n\nImproved tab completion for \"ALTER DEFAULT PRIVILEGE\" and \"ALTER TABLE\":\n1) GRANT, REVOKE and FOR USER keyword was not displayed in tab\ncompletion of alter default privileges like the below statement:\nALTER DEFAULT PRIVILEGES GRANT INSERT ON tables TO PUBLIC;\nALTER DEFAULT PRIV... |
[
{
"msg_contents": "Hi,\n\nI was rebasing my meson tree, which has more OSs added to CI, and noticed that\n010_database.pl started failing on openbsd recently-ish, without the CI\nenvironment for that having changed.\n\nThe tests passed on openbsd when my tree was based on 47b7051bc82\n(2023-06-01), but failed a... |
[
{
"msg_contents": "Hi,\n\nWhen looking at Assert() failures and at PANICs, the number of \"pointless\"\nstack entries at the top seems to have grown over the years. Here's an\nexample of a stacktrace (that I obviously intentionally triggered):\n\nProgram terminated with signal SIGABRT, Aborted.\n#0 __pthread_... |
[
{
"msg_contents": "Hi,\n\nI like that we now have a builtin backtrace ability. Unfortunately I think the\nbacktraces are often not very useful, because only externally visible\nfunctions are symbolized.\n\nE.g.:\n\n2023-07-02 10:54:01.756 PDT [1398494][client backend][:0][[unknown]] LOG: will crash\n2023-07-02... |
[
{
"msg_contents": "Hi,\n\nCommit 89e46da5e5 allowed us to use indexes for searching on REPLICA\nIDENTITY FULL tables. The documentation explains:\n\nWhen replica identity <quote>full</quote> is specified,\nindexes can be used on the subscriber side for searching the rows. Candidate\nindexes must be btree, non-... |
[
{
"msg_contents": "Hi all,\n\nWhile testing PG16, I observed that in PG16 there is a big performance\ndegradation in concurrent COPY into a single relation with 2 - 16\nclients in my environment. I've attached a test script that measures\nthe execution time of COPYing 5GB data in total to the single relation\nw... |
[
{
"msg_contents": "Hi hackers,\nStartup process will record not existed pages into hash table invalid_page_tab during replaying WAL.\nAnd it would call XLogCheckInvalidPages after reaching consistent recovery state. Finally, it will\nPANIC or WARNING based on parameter ignore_invalid_pages if where's any invali... |
[
{
"msg_contents": "Hi All,\nlogicalrep_message_type() is used to convert logical message type code\ninto name while prepared error context or details. Thus when this\nfunction is called probably an ERROR is already raised. If\nlogicalrep_message_type() gets an unknown message type, it will throw\nan error, whic... |
[
{
"msg_contents": "Normally it doesn't really matter which dbname is used in the connection\nstring that pg_basebackup and other physical replication CLI tools use.\nThe reason being, that physical replication does not work at the\ndatabase level, but instead at the server level. So you will always get\nthe dat... |
[
{
"msg_contents": "Hi,\n\nI noticed eelpout failed with this in stats.sql, in the pg_stat_io tests\nadded a couple months ago [1]:\n\n@@ -1415,7 +1415,7 @@\n :io_sum_vac_strategy_after_reuses >\n:io_sum_vac_strategy_before_reuses;\n ?column? | ?column?\n ----------+----------\n- t | t\n+ t ... |
[
{
"msg_contents": "Greetings,\n\nFor ISO and German dates the order DMY is completely ignored on output but\nused for input.\n\ntest=# set datestyle to 'ISO,DMY';\nSET\nselect '7-8-2023'::date\ntest-# ;\n date\n------------\n 2023-08-07\n(1 row)\n\ntest=# set datestyle to 'ISO,MDY';\nSET\ntest=# select '7-8-... |
[
{
"msg_contents": "Hi,\n\nI've run into an issue of a name clash with system libraries. Specifically,\nthe `Size` type seems to be just an alias for `size_t` and, at least on\nmacOS, it clashes with the core SDK, as it is also defined by MacTypes.h,\nwhich is used by some of the libraries one may want to use fr... |
[
{
"msg_contents": "At PgCon 2023 in Ottawa we had an Unconference session on Table Access\nMethods [1]\n\nOne thing that was briefly mentioned (but is missing from the notes)\nis need to have a sample API client in contrib/ , both for having a\n2nd user for API to make it more likely that non-heap AMs are doabl... |
[
{
"msg_contents": "Hello hackers,\n\nMy colleague, Ashwin, pointed out to me that brininsert's per-tuple init\nof the revmap access struct can have non-trivial overhead.\n\nTurns out he is right. We are saving 24 bytes of memory per-call for\nthe access struct, and a bit on buffer/locking overhead, with the\nat... |
[
{
"msg_contents": "Hello,\n\nThis is the first time I've run pgindent on my current machine, and it\ndoesn't seem to be making any modifications to source files. For\nexample this command:\n\n./src/tools/pgindent/pgindent src/backend/optimizer/path/allpaths.c\n\nleaves the allpaths.c file unchanged despite my h... |
[
{
"msg_contents": "Hello!\r\n\r\nI propose the attached patch to be applied on the 'master' branch\r\nof PostgreSQL to add GitLab CI automation alongside Cirrus CI in the PostgreSQL repository.\r\n\r\nIt is not intended to be a replacement for Cirrus CI, but simply suggestion for the\r\nPostgreSQL project to ho... |
[
{
"msg_contents": "Hi,\n\nAs discussed in [1], outputs of --help for some commands fits into 80 \ncolumns\nper line, while others do not.\n\nSince it seems preferable to have consistent line break policy and some \npeople\nuse 80-column terminal, wouldn't it be better to make all commands in 80\ncolumns per lin... |
[
{
"msg_contents": "Generate automatically code and documentation related to wait events\n\nThe documentation and the code is generated automatically from a new\nfile called wait_event_names.txt, formatted in sections dedicated to\neach wait event class (Timeout, Lock, IO, etc.) with three tab-separated\nfields:... |
[
{
"msg_contents": "Hi all,\n\nAfter removing --with-openssl from its build of HEAD, snapper has\nbegun failing in the pg_upgrade path 11->HEAD, because it attempts\npg_upgrade from binaries that have OpenSSL to builds without it:\nhttps://buildfarm.postgresql.org/cgi-bin/show_history.pl?nm=snapper&br=HEAD\n\nUs... |
[
{
"msg_contents": "Hi,\n\nI have a ROW LEVEL SECURITY policy on the table part of an extension, and\nwhile\ndumping the database where that extension is installed, dumps the policy of\nthat table too even though not dumpling that table .\n\nHere is quick tests, where I have added following SQL to adminpack--1.0... |
[
{
"msg_contents": "Hi,\n\nCurrently, there is no support for CHECK constraint DEFERRABLE in a create\ntable statement.\nSQL standard specifies that CHECK constraint can be defined as DEFERRABLE.\n\nThe attached patch is having implementation for CHECK constraint Deferrable\nas below:\n\n‘postgres[579453]=#’CREA... |
[
{
"msg_contents": "Hi,\n\nHeap-Only Tuple (HOT) updates are a significant performance\nenhancement, as they prevent unnecessary page writes. However, HOT\ncomes with a caveat: it means that if we have lots of available space\nearlier on in the relation, it can only be used for new tuples or in\ncases where ther... |
[
{
"msg_contents": "Hi,\n\n(patch proposal below).\n\nConsider a table with a FK pointing to a partitioned table.\n\n CREATE TABLE p ( id bigint PRIMARY KEY )\n PARTITION BY list (id);\n CREATE TABLE p_1 PARTITION OF p FOR VALUES IN (1);\n\n CREATE TABLE r_1 (\n id bigint PRIMARY KEY,\n p_id bigint... |
[
{
"msg_contents": "Hi, all. I want to report a bug about recovery of 2pc data, in current implementation of crash recovery, there are two ways to recover 2pc data:\n1、before redo, func restoreTwoPhaseData() will restore 2pc data those xid < ShmemVariableCache->nextXid, which is initialized from checkPoint.nextX... |
[
{
"msg_contents": "Hi.\n\nIn:\nhttps://git.postgresql.org/cgit/postgresql.git/tree/contrib/pg_freespacemap/pg_freespacemap.c\n\nrel = relation_open(relid, AccessShareLock);\n\nif (blkno < 0 || blkno > MaxBlockNumber)\nereport(ERROR,\n(errcode(ERRCODE_INVALID_PARAMETER_VALUE),\nerrmsg(\"invalid block number\")))... |
[
{
"msg_contents": "Hi hackers,\n\nEkaterina Sokolova and I have found a bug in PG 15. Since 88103567cb\nMarkGUCPrefixReserved() is supposed not only reporting GUCs that haven't\nbeen\ndefined by the extension, but also removing them. However, it removes them\nin\na wrong way, so that a GUC that goes after the r... |
[
{
"msg_contents": "Hi,\n\nWith PG16's 19d8e230, we got rid of BRIN's blocking of HOT updates,\nbut I just realized that we failed to update the README.HOT document\nwith this new exception for summarizing indexes.\n\nAttached a patch that updates that document, detailing the related rationale.\n\nI'm not sure i... |
[
{
"msg_contents": "We are now a few days in on Commitfest 2023-07, so it seems about time to send\nthe (by now) customary statistics email on how we are doing, and where we\nideally want go.\n\nThere are 350 patches registered in this commitfest, with 150 of those having\nbeen moved from the past commitfest. I... |
[
{
"msg_contents": "Hi,\n\nSimpleHash.\n\nThe function SH_START_ITERATE can trigger some overflow.\n\nSee:\ntypedef struct SH_ITERATOR\n{\nuint32 cur; /* current element */\nuint32 end;\nbool done; /* iterator exhausted? */\n} SH_ITERATOR;\n\nThe cur field is uint32 size and currently can be stored a uint64,\nwh... |
[
{
"msg_contents": "Thanks a lot Mark,\n\nI will take a look at this and get back to you if I find anything unclear\n\n---\nHannu\n\nOn Tue, Jul 4, 2023 at 10:14 PM Mark Dilger\n<mark.dilger@enterprisedb.com> wrote:\n>\n> Hackers,\n>\n> Over in [1], Hannu Krosing asked me to create and post several Table Access ... |
[
{
"msg_contents": "The following bug has been logged on the website:\n\nBug reference: 18016\nLogged by: Richard Vesely\nEmail address: richard.vesely@softea.sk\nPostgreSQL version: 15.3\nOperating system: Windows 10 Enterprise 22H2\nDescription: \n\nHi,\r\n\r\nGiven a table with a T... |
[
{
"msg_contents": "Windows has support for some signals[0], like SIGTERM and SIGINT. SIGINT\nmust be handled with care on Windows since it is handled in a separate\nthread. SIGTERM however can be handled in a similar way to UNIX-like\nsystems. I audited a few pqsignal calls that were blocked by WIN32 to\nsee if... |
[
{
"msg_contents": "Hi, hackers\n\nWhen I try to change log_destination using ALTER SYSTEM with the wrong value,\nit complains of the \"Unrecognized key word\" without available values. This\npatch tries to add a hint message that provides available values for\nlog_destination. Any thoughts?\n\n-- \nRegrads,\n... |
[
{
"msg_contents": "Hi, hackers!\n\nWhile analyzing -Wclobbered warnings from gcc we found a true one in \nPostgresMain():\n\npostgres.c: In function ‘PostgresMain’:\npostgres.c:4118:25: warning: variable \n‘idle_in_transaction_timeout_enabled’ might be clobbered by ‘longjmp’ or \n‘vfork’ [-Wclobbered]\n 4118 |... |
[
{
"msg_contents": "Hi,\n\nThis has already been discussed in [1].\nBut I thought it best to start a new thread.\n\nThe commit 31966b1\n<https://github.com/postgres/postgres/commit/31966b151e6ab7a6284deab6e8fe5faddaf2ae4c>\nintroduced the infrastructure to extend\nbuffers.\nBut the patch mixed types with int and... |
[
{
"msg_contents": "Dear hackers,\r\n\r\nThis is a fork thread from [1]. While analyzing codes I noticed that UPDATE and\r\nDELETE cannot be replicated when REPLICA IDENTITY is FULL and the table has datatype\r\nwhich does not have the operator class of Btree. I thnk this restriction is not\r\ndocumented but sho... |
[
{
"msg_contents": "Greetings, everyone!\n\nWhile working on an extension, I've found myself using valgrind on a \n32-bit OS (Debian 11)\nand after executing any query (even 'select 1;') under valgrind I've \nbeen shown the same error\neverytime:\n\n==00:00:00:18.109 2528== VALGRINDERROR-BEGIN\n==00:00:00:18.109... |
[
{
"msg_contents": "hello, all.\nRecently, I find one very strange situation to lose data of primary node which the\ndetails can be find at the first patch: 0001-Add-test-case-data-lost-after-restart.patch.\n\nThe first patch shows us that data could be lost after truncating physical file by\nsomeone else before... |
[
{
"msg_contents": "Hi hackers,\r\nI am learning the MemoryContext subsystem, but I really don't know where to find it's document (The PostgreSQL Document just provide some spi function).\r\nCan someone provide me some solutions?\r\nThanks in advance!\r\n\r\n\r\nYours,\r\nWen Yi\nHi hackers,I am learning t... |
[
{
"msg_contents": "Hi,\n\nJeff pointed out that one of the pg_stat_io tests has failed a few times\nover the past months (here on morepork [1] and more recently here on\nfrancolin [2]).\n\nFailing test diff for those who prefer not to scroll:\n\n+++ /home/bf/bf-build/francolin/HEAD/pgsql.build/testrun/recovery/... |
[
{
"msg_contents": "An instance compiled locally, without assertions, failed like this:\n\n< 2023-07-09 22:04:46.470 UTC >LOG: process 30002 detected deadlock while waiting for ShareLock on transaction 813219577 after 333.228 ms\n< 2023-07-09 22:04:46.470 UTC >DETAIL: Process holding the lock: 2103. Wait que... |
[
{
"msg_contents": "Hi, all.\nI want to report a bug about the recovery of two-phase transaction, in current implementation of crash recovery, there are two ways to recover 2pc data:\n1、before redo, func restoreTwoPhaseData() will restore 2pc data those xid < ShmemVariableCache->nextXid, which is initialized fro... |
[
{
"msg_contents": "Hello,\n\nWhile debugging one issue, we have added the below line in\npostgresGetForeignPlan() to see the foreignrel details.\n\n@@ -1238,6 +1238,8 @@ postgresGetForeignPlan(PlannerInfo *root,\n bool has_limit = false;\n ListCell *lc;\n\n+ elog(INFO, \"foreignrel: %s\", nodeT... |
[
{
"msg_contents": "Hi community,\r\nwhen I learn the source of PostgreSQL, I think it's better to add a tip to the postgres \"check mode\", this can help the postgres's user when they check the postgres's data directory.\r\n\r\n\r\n\r\nsrc/backend/bootstrap/bootstrap.c\r\n\r\n\r\n\r\nif (check_only)\r\n &n... |
[
{
"msg_contents": "Is $subject possible?\n\n I feel like maybe the answer is no, but then I can also see some backend\ncode for similar things in copy.h.\n\nPerhaps it’s possible via a function call not sending the SQL?\n\n- James\n\nIs $subject possible? I feel like maybe the answer is no, but then I can also ... |
[
{
"msg_contents": "This has been a long-standing annoyance of mine. Who hasn't done something\nlike this?:\n\npsql> SET random_page_cost = 2.5;\n(do some stuff, realize that rpc was too high)\n\nLet's put that inside of postgresql.conf:\n\n#-----------------------------------------------------------------------... |
[
{
"msg_contents": "Hi everyone,\n\nWe have been working on the pg_adviser\n<https://github.com/DrPostgres/pg_adviser> extension whose goal is to\nsuggest indexes by creating virtual/hypothetical indexes and see how it\naffects the query cost.\n\nThe hypothetical index shouldn't take any space on the disk (alloc... |
[
{
"msg_contents": "Hi,\nWhile testing some use cases, I encountered 'ERROR: attempted to update\ninvisible tuple' when a partitioned index is attached to a parent index\nwhich is also a replica identity index.\nBelow is the reproducible test case. The issue is seen only when the\ncommands are executed inside a... |
[
{
"msg_contents": "These patches were created during an unrelated discussion about pgbench.\n\nPlease see emails [1] - [6] linked below, for the past discussion.\n\nIn brief:\n\n> $ pgbench -i -I dtGvp -s 500\n\nThe init-steps are severely under-documented in pgbench --help output.\nI think at least a point... |
[
{
"msg_contents": "I realized that commit 19d8e2308bc5 (and 5753d4ee320b before that) added\na new output type to RelationGetIndexAttrBitmap but forgot to list its\neffect in the function's documenting comment. Here's a patch that\nupdates it, making it more specific (and IMO more readable). I also add\na com... |
[
{
"msg_contents": "Hi All,\nhttps://www.postgresql.org/docs/current/datatype-numeric.html gives me\n\"bad gateway\" error. Attached screen shot. Date/Time datatype\ndocumentation is accessible at\nhttps://www.postgresql.org/docs/current/datatype-datetime.html.\n\nJust got this while wrapping up for the day. Did... |
[
{
"msg_contents": "Greetings,\n\nWith a simple insert such as\n\nINSERT INTO test_table (cnt) VALUES (1), (2) RETURNING id\n\nif a portal is used to get the results then the CommandStatus is not\nreturned on the execute only when the portal is closed. After looking at\nthis more it is really after all of the da... |
[
{
"msg_contents": "Hi,\n\nI was reading the jit implementation and I notice that the lljit field of\nLLVMJitHandle is being assigned twice on llvm_compile_module function, is this\ncorrect? I'm attaching a supposed fixes that removes the second assignment. I\nran meson test and all tests have pass.\n\n--\nMath... |
[
{
"msg_contents": "We're observing a few cases with lockmanager spikes in a few quite loaded\nsystems.\n\nThese cases are different; queries are different, Postgres versions are 12,\n13, and 14.\n\nBut in all cases, servers are quite beefy (96-128 vCPUs, ~600-800 GiB)\nreceiving a lot of TPS (a few dozens of th... |
[
{
"msg_contents": "Hi,\n After I create the same name index on the heap table and the temporary\ntable, I can only get the temporary table's index by \\di+.\n\ncreate table t1(c1 int);\ncreate temp table t2(c1 int);\n\ncreate index idx1 on t1(c1);\n\\di+\n List of relations\n Schema | Na... |
[
{
"msg_contents": "Hi,\r\nI have a question about the routine \"GetNonHistoricCatalogSnapshot\".\r\nIt has a param \"Oid relid\". It firstly\r\nchecks if the relation has systemcache or if it is in \"RelationInvalidatesSnapshotsOnly\" related relations.\r\nIf yes, it will invalidate the CatalogSnapshot.\r\n... |
[
{
"msg_contents": "Hi,\n\nMy collegue Konstantin (cc-ed) noticed that the GiST code of intarray\nmay leak memory in certain index operations:\n\n> g_intbig_compress(...):\n> [...]\n> ArrayType *in = DatumGetArrayTypeP(entry->key);\n> [...]\n> if (in != DatumGetArrayTypeP(entry->key))\n> ... |
[
{
"msg_contents": "Hello, my name is Matheus Farias and this is the first time that I'm\nsending an email to the pgsql-hackers list. I'm a software developer intern\nat Bitnine Global Inc. and, along with other interns, we've been working on\nupdating Apache AGE with the latest version of Postgres, the REL_16_B... |
[
{
"msg_contents": "While adapting a Java implementation of the SQL parser, I noticed that \nin structures JsonArrayAgg, JsonArrayConstructor, \nJsonArrayQueryConstructor and JsonObjectConstrutor, the absent_on_null \nfield defaults to TRUE.\nBut in JsonObjectAgg, absent_on_null defaults to FALSE.\nIs that inten... |
[
{
"msg_contents": "Hi,\n\nWhile looking at [1] I started to wonder why it is safe that\nCreateCheckPoint() updates XLogCtl->RedoRecPtr after releasing the WAL\ninsertion lock:\n\n\t/*\n\t * Now we can release the WAL insertion locks, allowing other xacts to\n\t * proceed while we are flushing disk buffers.\n\t ... |
[
{
"msg_contents": "I looked into the performance gripe at [1] about pg_restore not making\neffective use of parallel workers when there are a lot of tables.\nI was able to reproduce that by dumping and parallel restoring 100K\ntables made according to this script:\n\ndo $$\nbegin\nfor i in 1..100000 loop\n exe... |
[
{
"msg_contents": "Hi hackers!\n\nWhile working on an extension I encountered a quite tricky question -\nthe extension (with functions in C) creates tables during function calls,\nthese tables must be protected from direct users' queries, at the same\ntime they must remain accessible for all functions of this e... |
[
{
"msg_contents": "Hi,\n\nSeveral loops which are important for query performance, like heapgetpage()'s\nloop over all tuples, have to call functions like\nHeapCheckForSerializableConflictOut() and PredicateLockTID() in every\niteration.\n\nWhen serializable is not in use, all those functions do is to to return... |
[
{
"msg_contents": "Hello hackers,\n\nAs a follow-up for the CVE-2023-2454 fix, I think that it makes sense to\ncompletely remove unsafe functions\nPushOverrideSearchPath()/PopOverrideSearchPath(), which are not used in the\ncore now.\nPlease look at the patch attached.\n\nBeside that, maybe it's worth to rename... |
[
{
"msg_contents": "Hi all,\n\nWhile scanning the code, I have noticed that a couple of code paths\nthat do syscache lookups are passing down directly Oids rather than\nDatums. I think that we'd better be consistent here, even if there is\nno actual bug.\n\nI have noticed 11 callers of SearchSysCache*() that pa... |
[
{
"msg_contents": "Hi all,\n\nI’m a security engineer and I’m looking into restricting the set of allowed ciphers on Postgres and configure a concrete set of curves on our postgres instances.\n\nI see in the source code that only TLS 1.2 and bellow cipher lists can be configured:\n\nhttps://github.com/postgres/... |
[
{
"msg_contents": "Hey list,\n\nI was working on a project with event triggers and was wondering if there\nwas any context from the developers around why some things make this list\nand others do not. Example: REVOKE/ GRANT are in the event trigger matrix\n[1] but REINDEX is not. Just wondering if there's a mai... |
[
{
"msg_contents": "Hi,\n\nIn a number of workloads one can see two wait events prominently:\nLWLock:WALWrite and LWLock:WALInsert. Unfortunately for both that is not very\ninformative:\n\nLWLock:WALWrite can be reported because there is genuine contention on the\nLWLock, or, more commonly, because several backe... |
[
{
"msg_contents": "Hi, hackers!\n\nI've tried sending this patch to community before, let me try it second\ntime. Patch is revised and improved compared to previous version.\n\nThis patch adds TOAST support for system tables pg_class,\npg_attribute and pg_largeobject_metadata, as they include ACL columns,\nwhic... |
[
{
"msg_contents": "Hi,\n\nContinuing a topic from earlier threads[1][2], I've been wondering\nabout how to de-klugify wal_sync_method=fsync_writethrough (a setting\nthat actually affects much more than just WAL), and how to do the\nright thing for our users on macOS and Windows by default. Commit\nd0c28601 was... |
[
{
"msg_contents": "---------- Forwarded message ---------\nFrom: Sahil Sojitra <sahilsojitra4555@gmail.com>\nDate: Tue, 18 Jul, 2023, 8:43 am\nSubject: Regarding Installation of PostgreSQL\nTo: <security@postgresql.org>\n\n\nHello Sir,\n I got stuck into an error repeatedly while installing\nPostg... |
[
{
"msg_contents": "Since posgres 13 there's the option to do a FORCE when dropping a database\n(so it disconnects current users) Documentation here:\nhttps://www.postgresql.org/docs/current/sql-dropdatabase.html\n\nI am currently using dir format for the output\n pg_dump -d \"bdname\" -F d -j 4 -v -f /tmp/dir... |
[
{
"msg_contents": "Hi hackers,\n\nWe have encountered an issue (invalid message length) when the\npassword length is > 1000 in pg 11,12,13 versions. This is due to the\nlimit(1000) on the max length of the password. In this case the\npassword is an access token(JWT) which can have varied lengths >\n1000. I see... |
[
{
"msg_contents": "Looking at the upgrade question in [0] made me realize that we discard\npotentially useful information for troubleshooting. When we check if the\ncluster is properly shut down we might as well include the status from\npg_controldata in the errormessage as per the trivial (but yet untested)\n... |
[
{
"msg_contents": "Dear pgsql:\n When we adding a custom system table and defining an index for it at the\nsame time, the code compilation is possible, but the following errors may\noccur when starting the database:\n\n -------------------------------------------------------------------------------------------... |
[
{
"msg_contents": "Dear pgsql:\n When we adding a custom system table and defining an index for it at the\nsame time, the code compilation is possible, but the following errors may\noccur when starting the database:\n\n -------------------------------------------------------------------------------------------... |
[
{
"msg_contents": "Hello.\n\nThere's an issue brought up in the -bugs list [1]. Since triggers are\ndeactivated on a subscriber by default, foreign key constraints don't\nfire for replicated changes. The docs state this is done to prevent\nrepetitive data propagation between tables on subscribers. But foreign\n... |
[
{
"msg_contents": "When looking at a patch in the CFBot I realized that the SSL tests generate\nbackend warnings under ENFORCE_REGRESSION_TEST_NAME_RESTRICTIONS due to roles\nand databases not following the regression test naming convention. While not\nimpacting the tested functionality, it's pretty silly to h... |
[
{
"msg_contents": "Hello postgres hackers,\n Recently I encountered an issue: pg_rewind fails when dealing with in-place tablespace. The problem seems to be that pg_rewind is treating in-place tablespace as symbolic link, while in fact it should be treated as directory.\n Here is the output of pg_rewind:\npg_re... |
[
{
"msg_contents": "Hi,\n\nWhile I'm working on the thread[1], I found that the function of\nworker_spi module fails if 'shared_preload_libraries' doesn't have\nworker_spi.\n\nThe reason is that the database name is NULL because the database name\nis initialized only when process_shared_preload_libraries_in_prog... |
[
{
"msg_contents": "Hi,\ni had a issue here, When executing a SELECT statement using an index-only scan, i got a wrong row number because there may be an inconsistency between the VM page visibility status and the visibility status of the page,the VM bit is set and page-level is clear\n\n\ni read the code and no... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.