Re: UdmSearch: Log Files
Paul Stewart wrote: What happens in cache db mode if you don't delete the log files but run an update with new information from the indexer.conf file? Is the system aware enough on the contents of the logs to add sites on without creating duplicate entries and/or remove stuff that's in the logs and that needs to be removed from the cache as well? I guess my question is: Does it matter if you delete the log files each time you run splitter? You can either delete log files or keep them. Splitter is smart enough and it will not create any duplicates. If some document presents twice (or more times) in log file, splitter will take newest version. __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
Re: UdmSearch: New message on the WebBoard #1: Segmentation Fault
Thomas Yengst wrote: Alexander Barkov wrote: Author: Alexander Barkov Email: [EMAIL PROTECTED] Message: It is a bug in HTDB, you may try to increase MaxDocSize in indexer.conf. This works - I have 100,000 rows in a table and increasing MaxDocSize by a factor of 10 makes indexer not seg-fault. Yes, it should work. It is old known bug and we'll fix it sometime. We hasn't already done it just because there are some difficalties to make it in the current programm structure and not so many people use HTDB. So, it is low priority task. __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
UdmSearch: I think i discovered a bug in boolean search
I am using MnoGoSearch 3.1.8 and a slightly modified version of PHP frontend UDMSearch-php-3.1.1.2. While searching using the "|" operator or by using "any match" i ran into a strange bug: if i search for 2 or more words then one word in the query appears to generate more results than it should. Thinking this was a bug in the PHP frontend i search for the same words using http://udm.aspseek.com/cgi-bin/search.cgi. But it happened again. Here are the strange results: Search for: cooling reactor http Search results: cooling: 4396 reactor: 1645 http: 4358918 Displaying documents 1-20 of total 4358629 found. You notice that for the word "http" it shows 4,358,918 results but only 4,358,629 documents are really found. Can you do something about it? __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
Re: UdmSearch: I think i discovered a bug in boolean search
Div wrote: I am using MnoGoSearch 3.1.8 and a slightly modified version of PHP frontend UDMSearch-php-3.1.1.2. While searching using the "|" operator or by using "any match" i ran into a strange bug: if i search for 2 or more words then one word in the query appears to generate more results than it should. Thinking this was a bug in the PHP frontend i search for the same words using http://udm.aspseek.com/cgi-bin/search.cgi. But it happened again. Here are the strange results: Search for: cooling reactor http Search results: cooling: 4396 reactor: 1645 http: 4358918 Displaying documents 1-20 of total 4358629 found. You notice that for the word "http" it shows 4,358,918 results but only 4,358,629 documents are really found. Example on udm.aspseek.com does not have boolean search. There an old version is installed without boolean search. __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
Re: UdmSearch: I think i discovered a bug in boolean search
On Fri 03 Nov 2000 09:27, you wrote: Div wrote: I am using MnoGoSearch 3.1.8 and a slightly modified version of PHP frontend UDMSearch-php-3.1.1.2. While searching using the "|" operator or by using "any match" i ran into a strange bug: if i search for 2 or more words then one word in the query appears to generate more results than it should. Thinking this was a bug in the PHP frontend i search for the same words using http://udm.aspseek.com/cgi-bin/search.cgi. But it happened again. Here are the strange results: Search for: cooling reactor http Search results: cooling: 4396 reactor: 1645 http: 4358918 Displaying documents 1-20 of total 4358629 found. You notice that for the word "http" it shows 4,358,918 results but only 4,358,629 documents are really found. Example on udm.aspseek.com does not have boolean search. There an old version is installed without boolean search. Yes, you're right. On udm.aspseek.com i searched using "any match". Anyway, the results are still wrong! __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
Re: UdmSearch: I think i discovered a bug in boolean search
Example on udm.aspseek.com does not have boolean search. There an old version is installed without boolean search. Yes, you're right. On udm.aspseek.com i searched using "any match". Anyway, the results are still wrong! All/any does not work too. __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
Re: UdmSearch: I think i discovered a bug in boolean search
Well, i don't give up! :) I searched on udm.aspseek.com only for http and it says: Search for: http Search results: http: 4358918 Displaying documents 1-20 of total 4358622 found. We've installed very beta version on askseek.com almost a half a year ago. So, very strange things may happen there. Why the "search results" number is higher than the "documents found" number? I remembered how i searched using boolean expressions on udm.aspseek.com: Search for: (http | web) Search results: http: 4358918 web: 648289 Displaying documents 1-20 of total 4358897 found. The same thing happened to me on the PHP frontend. Could you please help me? This seems to be a bug in PHP front-end. Sergey, could you help please? __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
Re[2]: UdmSearch: I think i discovered a bug in boolean search
Hi! Friday, November 03, 2000, 12:42:53 PM, you wrote: Search for: (http | web) Search results: http: 4358918 web: 648289 Displaying documents 1-20 of total 4358897 found. The same thing happened to me on the PHP frontend. Could you please help me? AB This seems to be a bug in PHP front-end. Sergey, could you help please? I need debug output from search.php. Cound you please give it to me ? -- Regards, Sergey aka gluke. __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
Re: Re[2]: UdmSearch: I think i discovered a bug in boolean search
On Fri 03 Nov 2000 10:49, Sergey Kartashoff wrote: Hi! Friday, November 03, 2000, 12:42:53 PM, you wrote: Search for: (http | web) Search results: http: 4358918 web: 648289 Displaying documents 1-20 of total 4358897 found. The same thing happened to me on the PHP frontend. Could you please help me? AB This seems to be a bug in PHP front-end. Sergey, could you help please? I need debug output from search.php. Cound you please give it to me ? I think the most eloquent example is: Begin ParseQ(): q=page | page | page Begin ParseStr(): qwe=page | page | page End ParseStr(): qwe=page|page|page End ParseQ(): q=( page || page || page ) load_cs(): SELECT path,link,name FROM categories WHERE path LIKE '__' ORDER BY NAME ASC load_cp(): SELECT name FROM categories WHERE path='' OR path IS NULL is_stopword(): SELECT count(*) FROM stopwor WHERE word='page' is_stopword(): SELECT count(*) FROM stopword WHERE word='page' is_stopword(): SELECT count(*) FROM stopword WHERE word='page' last_parse(): CREATE /*!32302 TEMPORARY */ TABLE t97325615967651800 ( url_id INT DEFAULT '0' NOT NULL, word_id INT DEFAULT '0' NOT NULL, intag TINYINT DEFAULT '0' NOT NULL, KEY i1t97325615967651800(url_id), KEY i2t97325615967651800(word_id)) last_parse(): INSERT INTO t97325615967651800 SELECT url_id,word_id,intag FROM ndict4 WHERE word_id = 336246304 last_parse(): SELECT count(*) FROM t97325615967651800 last_parse(): INSERT INTO t97325615967651800 SELECT url_id,word_id,intag FROM ndict4 WHERE word_id = 336246304 last_parse(): SELECT count(*) FROM t97325615967651800 last_parse(): INSERT INTO t97325615967651800 SELECT url_id,word_id,intag FROM ndict4 WHERE word_id = 336246304 last_parse(): SELECT count(*) FROM t97325615967651800 main(): SELECT SQL_SMALL_RESULT url_id, sum(intag) as r FROM t97325615967651800 GROUP BY url_id HAVING (( ( sum(word_id=336246304)0 ) OR ( sum(word_id=336246304)0 ) OR ( sum(word_id=336246304)0 ) )) ORDER BY r DESC Search results: page: 399; page: 399; page: 798; Documents 1-10 of 399 pages found. SearchTime: 1.15s. drop_temp_table(): DROP TABLE t97325615967651800 __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
Re[6]: UdmSearch: I think i discovered a bug in boolean search
Hi! Friday, November 03, 2000, 3:37:11 PM, you wrote: D I choose that senseless query to show that for the third word the number of D results are incorect. D Here is another example (you notice that the number of results for "http" is D equal to 1722+399, where 1722 is the corect one and 399 coresponds to "page"): ok , thank you. Could you please give me results of this queries : SELECT count(*) FROM ndict4 WHERE word_id = 410646757 SELECT count(*) FROM ndict4 WHERE word_id = 336246304 SELECT count(*) FROM ndict4 WHERE word_id = -1753739854 -- Regards, Sergey aka gluke. __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
UdmSearch: New message on the WebBoard #1: Segmentation Fault - core dumped after indexer
Author: Vlad Gerasimov Email: [EMAIL PROTECTED] Message: Hello! version: mnogosearch-3.1.8 with mysql After indexer I got: ... [1] http://www.novgorod.ru/city/culture/fufin/photo/22.htm [1] http://www.novgorod.ru/city/culture/fufin/photo/23.htm [1] http://www.novgorod.ru/city/culture/fufin/photo/24.htm [1] Done (1097 seconds) Segmentation Fault - core dumped It's bug or feature or what?? Thanks in advance, Vlad Reply: http://search.mnogo.ru/board/message.php?id=657 __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
Re[6]: UdmSearch: I think i discovered a bug in boolean search
Hi!
Friday, November 03, 2000, 3:45:17 PM, you wrote:
Dif ($row=fetch_row($res)) {
D $count=$row[0]-$count;
D $wordsinfo .= "b$old_norm_word/b: $count; ";
D $count=$row[0];
D}
ok, thank you! I will include this in the next release.
--
Regards, Sergey aka gluke.
__
If you want to unsubscribe send "unsubscribe udmsearch"
to [EMAIL PROTECTED]
UdmSearch: New message on the WebBoard #1: Segmentation Fault - core dumped after indexer
Author: Vlad Gerasimov Email: [EMAIL PROTECTED] Message: System: Solaris 7 x86 mysql: 3.23.21-beta Vlad Reply: http://search.mnogo.ru/board/message.php?id=658 __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
Re: UdmSearch: New message on the WebBoard #1: Segmentation Fault - core dumped after indexer
Hi! Friday, November 03, 2000, 1:47:08 PM, you wrote: VG After indexer I got: VG ... VG [1] http://www.novgorod.ru/city/culture/fufin/photo/22.htm VG [1] http://www.novgorod.ru/city/culture/fufin/photo/23.htm VG [1] http://www.novgorod.ru/city/culture/fufin/photo/24.htm VG [1] Done (1097 seconds) VG Segmentation Fault - core dumped VG It's bug or feature or what?? Please load indexer and ints coredump into gdb: gdb indexer core # backtrace and give us its output -- Regards, Sergey aka gluke. __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
UdmSearch: New message on the WebBoard #1: download timeout ?!
Author: Martin Perst Email: [EMAIL PROTECTED] Message: when my db reaches for example 50.000 links then this happens quite often when indexing; Indexer[27668]: [1] http://vsk.sk/banky/klbanky.cgi?id=6500 Indexer[27668]: [1] Download timeout Indexer[27668]: [1] http://rambo.elt.sk/minolta/minolta_t45_dealer.exe Indexer[27668]: [1] Download timeout Indexer[27668]: [1] http://www.minolta.sk/produkty/kopirovacie_stroje/di_151p.shtml Indexer[27668]: [1] Download timeout ReadTimeout is set to 90 When I try to access the given site using LYNX via shell then I get it always under 3-4 seconds. Where is the problem? Reply: http://search.mnogo.ru/board/message.php?id=660 __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
UdmSearch: New message on the WebBoard #1: Segmentation Fault - core dumped after indexer
Author: Vlad Gerasimov Email: [EMAIL PROTECTED] Message: backtrace: #0 0xdfb54806 in t_splay () from /usr/lib/libc.so.1 #1 0xdfb5470c in t_delete () from /usr/lib/libc.so.1 #2 0xdfb5447f in realfree () from /usr/lib/libc.so.1 #3 0xdfb549a3 in _free_unlocked () from /usr/lib/libc.so.1 #4 0xdfb5493c in free () from /usr/lib/libc.so.1 #5 0x805492a in UdmFreeStopList (Env=0x8080900) at sql.c:2144 #6 0x8067f78 in UdmFreeEnv (Env=0x8080900) at env.c:195 #7 0x804e66d in main (argc=1, argv=0x8047e18) at main.c:563 Reply: http://search.mnogo.ru/board/message.php?id=661 __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
UdmSearch: Some questions
I'm not getting much luck, here; I'm hoping if you could answer a couple of questions for me: 1) Security: I'm using the mySQL backend; how can I set up a user called udm_ro [or similar] so that if the password is stolen, they person who stole it can't delete any of the database. It's not possible just to remove create and drop permissions, because it uses temporary tables... Does anyone other than me even worry about this? 2) Tags: This is the best way to search through a selection of servers, right? What can I do in search.htm to make a search on two of the tags, but not any of the others? [I'm using the most up-to-date php front end] Thank-you very much, Gary (-; __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
UdmSearch: Strange error:
I just did a search for "chunky", and this happened: Search Time: 2.56s Search results: looking for: chunky; chunky: 26; Sorry, but search returned no results. Please try a different query. You can try to search this words: Argh! I'm using mySQL, and the PHP front-end Gary (-; __ If you want to unsubscribe send "unsubscribe udmsearch" to [EMAIL PROTECTED]
