Ultrasearch doesn't index documents processed by remote crawler

Question

Hi,
My Oracle9i 9.2 database is on Solaris. I have a remote crawler on Windows 2000. The remote crawler seems working fine. No error message is in log file. Every file has been processed. However, I can't query the documents processed by remote crawler. $ORACLE_HOME/ctx/log/ultrasearch_log reads:
Oracle Text, 9.2.0.1.0
15:24:45 07/12/02 begin logging
15:31:04 07/12/02 sync index: LING.WK$DOC_PATH_IDX
15:31:04 07/12/02 Begin document indexing
15:31:05 07/12/02 End of document indexing. 0 documents indexed.
The last part of the log file is:
=================== Crawling results ===================^M
Crawling started at 7/12/02 3:25 PM^M
Crawling stopped at 7/12/02 3:32 PM^M
Total crawling time = 0:6:31^M
^M
Total number of documents fetched = 179^M
Document fetch failures = 0^M
Document conversion failures = 0^M
Total number of unique documents indexed = 178^M
Total data collected = 1,975,751 bytes^M
Total number of non-indexable documents = 0^M
Average size of fetched document = 11,099 bytes^M
^M
Total indexing time = 0:0:0 for 1,975,751 bytes of data^M
Number of documents collected/indexed per hour = 1,638^M
^M
Number of times disk cache is full = 0^M
I have another crawler on the database host. It works fine. I can query the documents processed by this crawler.
Any idea?
Ling Niu

Answer 1

More Information about my question.
I used samba to share a directory /ling on Solaris to my Windows 2000, and map it to a drive on Windows NT, say, E:. This directory is used as both log and temp directory. I have a account on Solaris with same name/password as the my Widnows 2000 account. When the schedule is executing, I can see crawler create a directory inso_tmp on shared directory, and I believe it is used to filter. To my knowledge, after filtering, Oracle or crawler will copy the filtered file in inso_tmp to temp directory, which is /ling or E:\ in my case. But I failed to catch the temporary file which I know is transient files. I've given the write privilege on /ling to Oracle. I checked the table WK$DOC and WK$URL. In these tables, my temporary file's name is E:\****. If Oracle Server use this table to get the name of temporary file, it will fail because Oracle on Solaris doesn't know where is E:\. I can't get any log message to prove my guess. And if this is the case, it will be difficult to set up a remote crawler on Windows/Unix mixed enviroment, Right?
Any leads would be welcome,
Ling
Hi,
My Oracle9i 9.2 database is on Solaris. I have a remote crawler on Windows 2000. The remote crawler seems working fine. No error message is in log file. Every file has been processed. However, I can't query the documents processed by remote crawler. $ORACLE_HOME/ctx/log/ultrasearch_log reads:
Oracle Text, 9.2.0.1.0
15:24:45 07/12/02 begin logging
15:31:04 07/12/02 sync index: LING.WK$DOC_PATH_IDX
15:31:04 07/12/02 Begin document indexing
15:31:05 07/12/02 End of document indexing. 0 documents indexed.
The last part of the log file is:
=================== Crawling results ===================^M
Crawling started at 7/12/02 3:25 PM^M
Crawling stopped at 7/12/02 3:32 PM^M
Total crawling time = 0:6:31^M
^M
Total number of documents fetched = 179^M
Document fetch failures = 0^M
Document conversion failures = 0^M
Total number of unique documents indexed = 178^M
Total data collected = 1,975,751 bytes^M
Total number of non-indexable documents = 0^M
Average size of fetched document = 11,099 bytes^M
^M
Total indexing time = 0:0:0 for 1,975,751 bytes of data^M
Number of documents collected/indexed per hour = 1,638^M
^M
Number of times disk cache is full = 0^M
I have another crawler on the database host. It works fine. I can query the documents processed by this crawler.
Any idea?
Ling Niu