Re: [htdig-dev] Compressor::get_vals (fwd)

[Neal R. <ne...@ri...>]

  Has anyone else seen this error in the current snapshots?

  What executable does it happen on?

  I get it on htdig (the libhtdig version)	

  This error could happen on htmerge & htpurge.. any executable that is
changing the wordDB could corrupt the DBfile (if that's the problem).

Thanks

-- 
Neal Richter 
Knowledgebase Developer
RightNow Technologies, Inc.
Customer Service for Every Web Site
Office: 406-522-1485


---------- Forwarded message ----------
Date: Thu, 8 Aug 2002 13:47:58 -0500
From: And...@wi...
To: Neal Richter <ne...@ri...>
Subject: Re: [htdig-dev] Compressor::get_vals

You must have more debug in there, I don't see Exporting Document, nor can 
I find it in the code 07/28, 08/04 or mifluz 08/04 snapshots.  I did add 
the CDB debug and now it doesn't die ...

a

Andy Bach, Sys. Mangler
Internet: and...@wi... 
VOICE: (608) 261-5738  FAX 264-5030

"To understand recursion, we must first understand recursion."




Neal Richter <ne...@ri...>
08/08/02 01:11 PM

 
        To:     And...@wi...
        cc:     htd...@li...
        Subject:        Re: [htdig-dev] Compressor::get_vals


Andy,
                 Please try this for me:

Add this line to db/mp_cmpr.c before the call to cmpr_info->uncompress
printf("CDB___memp_cmpr_read: uncompress page [%d] \n", first_pgno);

Add this line to db/mp_cmpr.c before the call to cmpr_info->compress
printf("CDB___memp_cmpr_write: compress page [%d] \n", db_io->pgno);

Then run your htdig again and post the output.

Here's the output I get:

Exporting Document [010802-002096] [99955]
Exporting Document [010802-002106] [99956]
Exporting Document [010802-002109] [99957]
Exporting Document [010802-002119] [99958]
Exporting Document [010802-002122] [99959]
CDB___memp_cmpr_write: compress page [236291]
Exporting Document [010802-002145] [99960]
Exporting Document [010802-002190] [99961]
Exporting Document [010802-002213] [99962]
Exporting Document [010802-002223] [99963]
Exporting Document [010802-002229] [99964]
--
Exporting Document [010803-000314] [100195]
Exporting Document [010803-000332] [100196]
Exporting Document [010803-000353] [100197]
Exporting Document [010803-000360] [100198]
Exporting Document [010803-000361] [100199]
CDB___memp_cmpr_read: uncompress page [236291]
Exporting Document [010803-000362] [100200]
Exporting Document [010803-000363] [100201]
Exporting Document [010803-000364] [100202]
Exporting Document [010803-000365] [100203]
Exporting Document [010803-000367] [100204]
--
Exporting Document [010804-001475] [101212]
Exporting Document [010804-001476] [101213]
Exporting Document [010804-001477] [101214]
Exporting Document [010804-001478] [101215]
Exporting Document [010804-001479] [101216]
CDB___memp_cmpr_read: uncompress page [236291]


Note that the page is writen, read, and read again (it dumps on the
second read).  This is a bit unexpected, I would have expected a write
(with an error), then a read that dumps.

I am also noticing that it is somewhat inconsistent in when it dumps..

This leads me to believe its a memory corruption issue...  I'll run
Insure++ on it, and wait for it to finish ;-)

Thanks!


On Wed, 7 Aug 2002 And...@wi... wrote:

> Hey.
> 
> I was working w/ the 3.2.0b4 7/28 snapshot and hit:
> FATAL ERROR:Compressor::get_vals invalid comptype
> FATAL ERROR at file:WordBitCompress.cc line:827 !!!
> Segmentation Fault - core dumped
> 
> not sure where it was from (can't reproduce it) but I went to 
> htword/WordBitCompress.cc and noticed that, though the run was w/ -vvvvv 

> the:
>     if(verbose)printf("get_vals:comptype:%d\n",comptype);
> 
> stmts didn't show up in the output.  I can only find 2 places where new 
> Compressors are created, in htword/WordDBPage.cc and neithor of them 
seems 
> to pass along the verbose flag. I'm out of my depth here, but the 
> constructor for Compressors:
>    Compressor():BitStream()
>         {
>             verbose=0;
>         }
>     Compressor(int size0):BitStream(size0)
>         {
>             verbose=0;
>         }
> 
> seem to be setting always to zero.
> 
> a
> 
> Andy Bach, Sys. Mangler
> Internet: and...@wi... 
> VOICE: (608) 261-5738  FAX 264-5030
> 
> "To understand recursion, we must first understand recursion."
> 
> 
> -------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> 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