<me (AT) alternize (DOT) com> writes:
This is undoubtedly data-dependent. Can you supply some sample data
that makes it happen?

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 1: if posting/reading through Usenet, please send an appropriate
subscribe-nomail command to majordomo (AT) postgresql (DOT) org so that your
message can get through to the mailing list cleanly

it's not only happening with INSERTS, but also updates. as thats easier to
test, here's how i can reproduce the error:

1. create new database (encoding: UTF8) with tsearch2 on 8.2b1 win32 (system
locale: de_CH.1252)
2. insert the data from the zip file
[http://alternize.com/pgsql/tsearch2test.zip] (be sure to also update
pg_ts_cf /
pg_ts_cfgmap as we have WIN1252 locale)
3. execute UPDATE test SET idxFTI = to_tsvector('default', "sometext"); or
similar queries
4. hopefully see the process crashing as i do ;-)


2006-10-17 17:23:44 LOG: server process (PID 4584) exited with exit
code -1073741819
2006-10-17 17:23:44 LOG: terminating any other active server processes
2006-10-17 17:23:44 WARNING: terminating connection because of crash of
another server process
2006-10-17 17:23:44 DETAIL: The postmaster has commanded this server
process to roll back the current transaction and exit, because another
server process exited abnormally and possibly corrupted shared memory.
{snipp}
2006-10-17 17:23:44 LOG: all server processes terminated; reinitializing
2006-10-17 17:23:44 LOG: database system was interrupted at 2006-10-17
17:23:41 W. Europe Daylight Time
2006-10-17 17:23:44 LOG: Windows fopen("recovery.conf","r") failed: code 2,
errno 2
2006-10-17 17:23:44 LOG: Windows fopen("pg_xlog/00000001.history","r")
failed: code 2, errno 2
2006-10-17 17:23:44 LOG: Windows fopen("backup_label","r") failed: code 2,
errno 2
2006-10-17 17:23:44 LOG: checkpoint record is at 0/E2ECA728
2006-10-17 17:23:44 LOG: redo record is at 0/E2ECA728; undo record is at
0/0; shutdown FALSE
2006-10-17 17:23:44 LOG: next transaction ID: 0/514299; next OID: 6276957
2006-10-17 17:23:44 LOG: next MultiXactId: 1; next MultiXactOffset: 0
2006-10-17 17:23:44 LOG: database system was not properly shut down;
automatic recovery in progress
2006-10-17 17:23:44 LOG: redo starts at 0/E2ECA778
2006-10-17 17:23:44 LOG: unexpected pageaddr 0/DB0CC000 in log file 0,
segment 227, offset 835584
2006-10-17 17:23:44 LOG: redo done at 0/E30CBE78
2006-10-17 17:23:45 LOG: database system is ready
2006-10-17 17:23:45 LOG: Windows fopen("global/pg_fsm.cache","rb") failed:
code 2, errno 2
2006-10-17 17:23:45 LOG: transaction ID wrap limit is 2147484172, limited
by database "postgres"
2006-10-17 17:23:45 LOG: Windows fopen("global/pgstat.stat","rb") failed:
code 2, errno 2


i've also tried to update each record on its own in a for-loop. here the
crash happens as well, sometimes after 10 updates, sometimes after 100
updates, sometimes even after 1 update. but eventually every record can be
updated. so i do not think its entierly content-related...

for what its worth, here's the output of pg_controldata:

pg_control version number: 822
Catalog version number: 200609181
Database system identifier: 4986650172201464825
Database cluster state: in production
pg_control last modified: 17.10.2006 17:44:29
Current log file ID: 0
Next log file segment: 230
Latest checkpoint location: 0/E4E0F978
Prior checkpoint location: 0/E46BF420
Latest checkpoint's REDO location: 0/E4E03098
Latest checkpoint's UNDO location: 0/0
Latest checkpoint's TimeLineID: 1
Latest checkpoint's NextXID: 0/531333
Latest checkpoint's NextOID: 6285149
Latest checkpoint's NextMultiXactId: 1
Latest checkpoint's NextMultiOffset: 0
Time of latest checkpoint: 17.10.2006 17:43:45
Minimum recovery ending location: 0/0
Maximum data alignment: 8
Database block size: 8192
Blocks per segment of large relation: 131072
WAL block size: 8192
Bytes per WAL segment: 16777216
Maximum length of identifiers: 64
Maximum columns in an index: 32
Date/time type storage: floating-point numbers
Maximum length of locale name: 128
LC_COLLATE: German_Switzerland.1252
LC_CTYPE: German_Switzerland.1252

let me know if more information / data is needed.

on a sidenote: are those fopen() errors debug-code-leftovers or something
one should worry about? i can't find those files on the file system.

- thomas



---------------------------(end of broadcast)---------------------------
TIP 4: Have you searched our list archives?

http://archives.postgresql.org

[offline ATM, can't look at the main issue, just wanted to get a comment
in about fopen(). Unless beaten to it, I'll try to look at the other
part later.]


No. It's debugging output for a problem we had earlier, and will be
removed before release.

//Magnus

---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match

thanks for clarifying. wasn't sure whether the comment in the release notes
really was applying to these "file not found" error messages.

in case more information / data is needed please let me know.
if helpfull (i.e. bug not reproducable with my test data in the previous
mail), full pgsql access could be set up as well.

thanks in advance,
thomas



---------------------------(end of broadcast)---------------------------
TIP 2: Don't 'kill -9' the postmaster

just a small update: this problem is also present in beta 2.
not a big problem for the moment, as we currently have disabled fulltext
search capabilities on the website.

regards,
thomas

----- Original Message -----
From: <me (AT) alternize (DOT) com>
To: "Tom Lane" <tgl (AT) sss (DOT) pgh.pa.us>
Cc: <pgsql-bugs (AT) postgresql (DOT) org>
Sent: Tuesday, October 17, 2006 10:19 PM
Subject: Re: [BUGS] 8.2beta1 (w32): server process crash (tsvector)




---------------------------(end of broadcast)---------------------------
TIP 5: don't forget to increase your free space map settings

this bug (please see below) is unfortunately still present in beta3 (win32
build). test case still crashes the child process and lets postmaster kill &
reload everything.

it is not GiST-related, i've just validated the same problem using GIN.

this breaks tsearch2 functionality on our win32 system as no
tsvector-indexing of new/existing rows is possible (crash after ~10
processed rows). searching already indexed rows works fine.

best regards,
thomas

----- Original Message -----
From: <me (AT) alternize (DOT) com>
To: "Tom Lane" <tgl (AT) sss (DOT) pgh.pa.us>
Cc: <pgsql-bugs (AT) postgresql (DOT) org>
Sent: Tuesday, October 17, 2006 9:19 PM
Subject: Re: [BUGS] 8.2beta1 (w32): server process crash (tsvector)




---------------------------(end of broadcast)---------------------------
TIP 6: explain analyze is your friend

Ok, I've run this test on an assert enabled build (my msvc build,
actually, so I could get a debugger on it if needed). It then outputs:

WARNING: detected write past chunk end in ExprContext 02C6D0F8
WARNING: detected write past chunk end in ExprContext 02C6AEA0
WARNING: detected write past chunk end in ExprContext 02C6B200
WARNING: detected write past chunk end in ExprContext 02C44630
WARNING: detected write past chunk end in ExprContext 02C4C118
WARNING: problem in alloc set ExprContext: bogus aset link in block
02C435A8, c
hunk 02C44520
WARNING: detected write past chunk end in ExprContext 02C66440
WARNING: detected write past chunk end in ExprContext 02C3B9D0
WARNING: detected write past chunk end in ExprContext 02C3BDE8
WARNING: detected write past chunk end in ExprContext 02C4E7E0
WARNING: detected write past chunk end in ExprContext 02C47508
WARNING: problem in alloc set ExprContext: bogus aset link in block
02C435A8, c
hunk 02C47528
WARNING: detected write past chunk end in ExprContext 02C43800
WARNING: detected write past chunk end in ExprContext 02C66C90
WARNING: detected write past chunk end in ExprContext 02C68270
WARNING: detected write past chunk end in ExprContext 02C4F5D8
WARNING: problem in alloc set ExprContext: bogus aset link in block
02C4E6F8, c
hunk 02C4F5F8
WARNING: detected write past chunk end in ExprContext 02C7B680
WARNING: detected write past chunk end in ExprContext 02C45190
WARNING: detected write past chunk end in ExprContext 02C46AC8
WARNING: detected write past chunk end in ExprContext 02C3C538
WARNING: detected write past chunk end in ExprContext 02C67B90
WARNING: detected write past chunk end in ExprContext 02C438F0
WARNING: problem in alloc set ExprContext: bad single-chunk 02C43FB8 in
block 0
2C435A8
WARNING: problem in alloc set ExprContext: bogus aset link in block
02C435A8, c
hunk 02C43FB8
server closed the connection unexpectedly
This probably means the server terminated abnormally
before or while processing the request.=20




The actual crash happens on line 1142 of AllocSetCheck. Full callstack
is:
postgres.exe!AllocSetReset(MemoryContextData *
context=3D0x02c455b8) Line 409 + 0x9 bytes C
postgres.exe!MemoryContextReset(MemoryContextData *
context=3D0x02c455b8) Line 129 + 0xf bytes C
postgres.exe!ExecScan(ScanState * node=3D0x02c227d0,
TupleTableSlot * (ScanState *)* accessMtd=3D0x00530b70) Line 91 + 0xc
bytes C
postgres.exe!ExecSeqScan(ScanState * node=3D0x02c227d0) Line 130
+ 0xe bytes C
postgres.exe!ExecProcNode(PlanState * node=3D0x02c227d0) Line 349
+ 0x9 bytes C
postgres.exe!ExecutePlan(EState * estate=3D0x02c223e8, PlanState *
planstate=3D0x02c227d0, CmdType operation=3DCMD_UPDATE, long numberTuples=
=3D0,
ScanDirection direction=3DForwardScanDirection, _DestReceiver *
dest=3D0x02bf8d30) Line 1081 + 0x9 bytes C
postgres.exe!ExecutorRun(QueryDesc * queryDesc=3D0x02c44078,
ScanDirection direction=3DForwardScanDirection, long count=3D0) Line 246 +
0x20 bytes C
postgres.exe!ProcessQuery(Query * parsetree=3D0x02bddab0, Plan *
plan=3D0x02bf7e28, ParamListInfoData * params=3D0x00000000, _DestReceiver *
dest=3D0x02bf8d30, char * completionTag=3D0x00d4fb24) Line 157 + 0xd bytes
C
postgres.exe!PortalRunMulti(PortalData * portal=3D0x02c48138,
_DestReceiver * dest=3D0x02bf8d30, _DestReceiver * altdest=3D0x02bf8d30,
char * completionTag=3D0x00d4fb24) Line 1148 + 0x1c bytes C
postgres.exe!PortalRun(PortalData * portal=3D0x02c48138, long
count=3D2147483647, _DestReceiver * dest=3D0x02bf8d30, _DestReceiver *
altdest=3D0x02bf8d30, char * completionTag=3D0x00d4fb24) Line 700 + 0x15
bytes C
postgres.exe!exec_simple_query(const char *
query_string=3D0x02bdd4c8) Line 943 + 0x23 bytes C
postgres.exe!PostgresMain(int argc=3D4, char * * argv=3D0x02ba2478,
const char * username=3D0x025e7af8) Line 3414 + 0xc bytes C
postgres.exe!BackendRun(Port * port=3D0x00d4fd2c) Line 2853 +
0x17 bytes C
postgres.exe!SubPostmasterMain(int argc=3D3, char * *
argv=3D0x025e5b30) Line 3288 + 0xc bytes C
postgres.exe!main(int argc=3D3, char * * argv=3D0x025e5b30) Line
165 + 0xd bytes C




Can someone point me towards what next to check? :-)

Oh, and per Thomas, this does NOT appear to happen in C locale, but
happens in both de_CH and en_US. ANd I've tested in Swedish_Sweden,
which also crashes.

Note that this appears not to be index related at all, because during
this test there is only a standard btree index on "someint", and nothing
on the tsvector column.

//Magnus


---------------------------(end of broadcast)---------------------------
TIP 3: Have you checked our extensive FAQ?

http://www.postgresql.org/docs/faq

yeah... i dont think it will be related to index anyway.
it looks to me that some extra bytes are written to the allocated
memory when locale is set to en-US etc. The warning

"WARNING: detected write past chunk end in ExprContext 02C6D0F8"

is generated when you write bytes on a location more than its allocated size,
and this thing will eventualy lead to a server crash.

Now, I would suggest try to minimize the script and look for the action
with leads to this warning. Then a debug would be easy on the execution
path and I guess, there won't be any problem in parsing and planning.

Can you send me the knotty script BTW?


--Imad
www.EnterpriseDB.com

On 11/12/06, Magnus Hagander <mha (AT) sollentuna (DOT) net> wrote:
---------------------------(end of broadcast)---------------------------
TIP 2: Don't 'kill -9' the postmaster

here are the steps to reproduce:

--------
1. intall win32 beta3 as new instance using UTF8 / en_US and tsearch2
module, everything else default
2. create new db with encoding UTF8, standard template
3. load data from http://alternize.com/pgsql/tsearch2test.zip (updated dump
so it only includes the test table/data)
4. issue query: UPDATE test SET idxFTI = to_tsvector('default', sometext);
5. watch the process die... *sniff*
-------

(steps 1 & 2 can probably be skipped, but i wanted to have a clean test env)

best regards,
thomas



----- Original Message -----
From: "imad" <immaad (AT) gmail (DOT) com>
To: "Magnus Hagander" <mha (AT) sollentuna (DOT) net>
Cc: "Thomas H." <me (AT) alternize (DOT) com>; <pgsql-bugs (AT) postgresql (DOT) org>; "Tom Lane"
<tgl (AT) sss (DOT) pgh.pa.us>
Sent: Sunday, November 12, 2006 5:20 PM
Subject: Re: [BUGS] 8.2beta1 (w32): server process crash (tsvector)




---------------------------(end of broadcast)---------------------------
TIP 5: don't forget to increase your free space map settings

"Thomas H." <me (AT) alternize (DOT) com> writes:
Did you do anything to install tsearch2 into this fresh database beyond
"\i tsearch2.sql"?

I couldn't reproduce it on a Linux x86 machine, so it seems specific to
the Windows port.

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 5: don't forget to increase your free space map settings