DURING REBUILDS
Rebuildtext
/Search/Config/Databases/KUNI.tbl tells you where all the keyword indexes are stored
Our *txt1.db is in /z/sirsi/Unicorn/Searchdb/KUNI
txt0.db is in /y....
*this was half the size after rebuild in 2001 -- went from over 1 GB to 555 MB
Selden says better to try tailing one of the intermediate files that the report creates as it runs. Or search for Total in DB. Grep for "Total in DB" in the rptXXXX file in /tmp that ends in 1
Combining the "total in DB" info with the time stamp that follows each time it finishes indexing a clump of records gives a good idea of how things are going. The process gradually slows down as the index grows larger and each reorg section takes longer and longer.
WRK1 and WRK2 specify where the intermediate work/temp files are stored for the BRS rebuild, which in our case is
/z/sirsi/Unicorn/Searchdb/KUNI for rebuildtext
Rebuildheading
/t/sirsi/Unicorn/Database
watch headingved.fil grow to larger than on test server
Rebuildauth
watch authindex.dat and .idx in
/load4/sirsi/Unicorn/Database
Alan Hagyard's trick for expediting rebuildtext
We did try this on our test server and it worked fine.
1) schedule rebuildtext as NEVER - enter an unusual # for the limit -- 249999
2) go to the Rptsched directory
3) grep 249999 *
chsb.selans: *# Titles per load:249999
chsb.set:rebuild|1|249999|249999|
4) edit the two files - change 249999 to 500000
(warning: do not go over 700000 here)
5) modify rebuildtext to run asap (or schedule for later)
You do need enough disk space for the larger work file than as if you chose 250000 records at a time.
Other info from Alan Hagyard
You can get a pretty good idea of how far along you are in a rebuildtext by grepping for "Total in DB" in the rptXXXX file in /tmp that ends in 1
If you go to the Work directory, you will see a number of files with the names "UNIxx" where xx goes from aa to az, ba to bz, etc. The number of files left equals the number of iterations to go in your rebuild. So, if you have 2 million records and use 250,00 you will have UNIaa through UNIah at the beginning (actually after about an hour or so). If you use 25,000 you will have UNIaa through UNIdb at the beginning (i.e. 80 files). Since each iteration does a reorg and the reorg takes half an hour (toward the end of my rebuild), you would be doing a lot of reorg in this example.