Menu
▾
▴
htdig-dev
|
From: Lachlan A. <lh...@us...> - 2003-02-27 12:44:24
|
Greetings all,
I've found a (much) smaller data set that gives me errors.
What do other people get with:
./configure --enable-tests
make
make check
make check
I get t_htmerge failing the second time, with a read error on page 3=20
(if I remove the redirection of stderr).
Cheers,
Lachlan
|
|
From: Neal R. <ne...@ri...> - 2003-02-27 19:12:47
|
Great! I'll test and fix this ASAP. On Thu, 27 Feb 2003, Lachlan Andrew wrote: > Greetings all, > > I've found a (much) smaller data set that gives me errors. > What do other people get with: > ./configure --enable-tests > make > make check > make check > > I get t_htmerge failing the second time, with a read error on page 3 > (if I remove the redirection of stderr). > > Cheers, > Lachlan > > > ------------------------------------------------------- > This SF.NET email is sponsored by: > SourceForge Enterprise Edition + IBM + LinuxWorld http://www.vasoftware.com > _______________________________________________ > htdig-dev mailing list > htd...@li... > https://lists.sourceforge.net/lists/listinfo/htdig-dev > Neal Richter Knowledgebase Developer RightNow Technologies, Inc. Customer Service for Every Web Site Office: 406-522-1485 |
|
From: Lachlan A. <lh...@us...> - 2003-03-05 11:08:47
|
Greetings Neal, Have you had any luck reproducing this bug? I've got as far as finding that, at some point, page 3 has a chain=20 3->71->34, but that page 27 is then allocated the chain 27->70->71,=20 so page 3 gets corrupted. I'm about to start looking for the=20 free-list. Cheers, Lachlan On Friday 28 February 2003 06:14, Neal Richter wrote: > Great! I'll test and fix this ASAP. |
|
From: Lachlan A. <lh...@us...> - 2003-03-05 11:52:29
|
Just following up on that, the problem seems to be that =20
dbmfp->mfp->last_pgno doesn't get read in properly when the database=20
is reopened. I *assume* that it should be the last page actually=20
used, but at some point the free list allocates 71 (which was used=20
in the database when it was written in the previous program) despite=20
dbmfp->mfp->last_pgno having been reset to 33. I added the if =20
statement below in the diagnostic section at the end of=20
CDB___memp_cmpr_alloc, and the only place it reports an error is the=20
one that causes the crash...
#ifdef DEBUG_CMPR
fprintf(stderr,"CDB___memp_cmpr_alloc:: reuse free page %d from=20
weakcmpr\n", *pgnop);
if (*pgnop > dbmfp->mfp->last_pgno)
fprintf (stderr, "*******ERROR?? dbmfp->mfp->last_pgno %d,\=20
allocating %d\n", dbmfp->mfp->last_pgno, *pgnop);
#endif
Good night :)
Lachlan
On Wednesday 05 March 2003 22:08, Lachlan Andrew wrote:
> I've got as far as finding that, at some point, page 3 has a chain
> 3->71->34, but that page 27 is then allocated the chain 27->70->71,
> so page 3 gets corrupted. I'm about to start looking for the
> free-list.
|
|
From: Jim C. <li...@yg...> - 2003-02-28 05:08:19
|
Hi - I can't currently duplicate this result; however make check is pretty much a disaster on both systems that I have tried (Red Hat 8.0 and OS X). On the Red Hat box, I receive all sorts of error messages regarding modules and about half of the tests end up failing. Perhaps this is due to the fact that I am running a 2.0.x version of Apache? I don't have time to dig into it at the moment. Under OS X, I can't even build all of the test programs; WordType::instance shows up as an undefined symbol when linking testnet Jim On Thursday, February 27, 2003, at 05:43 AM, Lachlan Andrew wrote: > Greetings all, > > I've found a (much) smaller data set that gives me errors. > What do other people get with: > ./configure --enable-tests > make > make check > make check > > I get t_htmerge failing the second time, with a read error on page 3 > (if I remove the redirection of stderr). > > Cheers, > Lachlan > > > ------------------------------------------------------- > This SF.NET email is sponsored by: > SourceForge Enterprise Edition + IBM + LinuxWorld Something 2 See! > http://www.vasoftware.com > _______________________________________________ > htdig-dev mailing list > htd...@li... > https://lists.sourceforge.net/lists/listinfo/htdig-dev |
|
From: Neal R. <ne...@ri...> - 2003-03-17 22:51:49
|
> I've found a (much) smaller data set that gives me errors. > What do other people get with: > ./configure --enable-tests > make > make check > make check > > I get t_htmerge failing the second time, with a read error on page 3 > (if I remove the redirection of stderr). Hmm. I can't get this to duplicate!!! ARG! What kind of Linux system do you have running? Thanks! Neal Richter Knowledgebase Developer RightNow Technologies, Inc. Customer Service for Every Web Site Office: 406-522-1485 |
|
From: Lachlan A. <lh...@us...> - 2003-03-22 06:22:03
|
Thanks for looking at this, Neal. I get the panic using Mandrake 8.2 with % gcc -v Reading specs from /usr/lib/gcc-lib/i586-mandrake-linux-gnu/3.2/specs Configured with: ../configure --prefix=3D/usr --libdir=3D/usr/lib=20 --with-slibdir=3D/lib --mandir=3D/usr/share/man --infodir=3D/usr/share/in= fo=20 --enable-shared --enable-threads=3Dposix --disable-checking=20 --enable-long-long --enable-__cxa_atexit=20 --enable-languages=3Dc,c++,ada,f77,objc,java=20 --host=3Di586-mandrake-linux-gnu --with-system-zlib Thread model: posix gcc version 3.2 (Mandrake Linux 9.0 3.2-1mdk) % locate libz.so /usr/lib/libz.so /usr/lib/libz.so.1.1.4 /lib/libz.so.1 /lib/libz.so.1.1.4 % ld -v GNU ld version 2.12.90.0.15 20020717 Andy Bach has also reported this problem. Andy, what system are you=20 using? When I put in debugging code to see what was happening, the error=20 started manifesting itself in t_htdig_local instead, but it was=20 still page 3. The problem is that the free list for _weakcmpr =20 seems to be getting corrupted. Page 3 overflows into a chain=20 3->71->72, but then another page overflows into page 71, corrupting=20 page 3. Do you know how to display the free list? I haven't even been able to=20 find where new pages are allocated -- every scent I follow leads back=20 to *memory* page allocation, rather than *file* page allocation :( =20 The file db_vrfy.c has code to print the free list, but the file=20 seems to have been copied in from a later version of BDB, and never=20 integrated properly. Thanks, Lachlan On Tuesday 18 March 2003 09:51, Neal Richter wrote: > > ./configure --enable-tests > > make > > make check > > make check > Hmm. I can't get this to duplicate!!! > ARG! What kind of Linux system do you have running? |
|
From: Neal R. <ne...@ri...> - 2003-03-22 17:12:13
|
I can't get duplication on Mandrake 8.1, RedHat 8.0, and RedHat 7.3 machines. The RedHat machines don't make it through ./configure --enable_tests, various errors happen. I'm reading the configure script to figure out why. The Mandrake 8.1 machine passes all tests repeatedly with no errors in htmerge. Thanks Neal. On Sat, 22 Mar 2003, Lachlan Andrew wrote: > Thanks for looking at this, Neal. > > I get the panic using Mandrake 8.2 with > > % gcc -v > Reading specs from /usr/lib/gcc-lib/i586-mandrake-linux-gnu/3.2/specs > Configured with: ../configure --prefix=/usr --libdir=/usr/lib > --with-slibdir=/lib --mandir=/usr/share/man --infodir=/usr/share/info > --enable-shared --enable-threads=posix --disable-checking > --enable-long-long --enable-__cxa_atexit > --enable-languages=c,c++,ada,f77,objc,java > --host=i586-mandrake-linux-gnu --with-system-zlib > Thread model: posix > gcc version 3.2 (Mandrake Linux 9.0 3.2-1mdk) > > % locate libz.so > /usr/lib/libz.so > /usr/lib/libz.so.1.1.4 > /lib/libz.so.1 > /lib/libz.so.1.1.4 > > % ld -v > GNU ld version 2.12.90.0.15 20020717 > > > Andy Bach has also reported this problem. Andy, what system are you > using? > > When I put in debugging code to see what was happening, the error > started manifesting itself in t_htdig_local instead, but it was > still page 3. The problem is that the free list for _weakcmpr > seems to be getting corrupted. Page 3 overflows into a chain > 3->71->72, but then another page overflows into page 71, corrupting > page 3. > > Do you know how to display the free list? I haven't even been able to > find where new pages are allocated -- every scent I follow leads back > to *memory* page allocation, rather than *file* page allocation :( > The file db_vrfy.c has code to print the free list, but the file > seems to have been copied in from a later version of BDB, and never > integrated properly. > > Thanks, > Lachlan > > On Tuesday 18 March 2003 09:51, Neal Richter wrote: > > > ./configure --enable-tests > > > make > > > make check > > > make check > > Hmm. I can't get this to duplicate!!! > > ARG! What kind of Linux system do you have running? > Neal Richter Knowledgebase Developer RightNow Technologies, Inc. Customer Service for Every Web Site Office: 406-522-1485 |
Thanks for helping keep SourceForge clean.
X