Okay, here's some specifics about the problem I'm experiencing: Alternate Key Index Details for File BIN.QUEUE Page 1
File.................. BIN.QUEUE Alternate key length.. 35 Node/Block size....... 4K OV blocks............. 1 (0 in use, 0 overflowed) Indices............... 1 (0 D-type) Index updates......... Enabled, No updates pending Index-Name...... F-type K-type Built Empties Dups In-DICT S/M F-no/VF-expr.... BTREE_DEFAULT V Txt Yes Yes Yes Yes M OCONV(@ID,"G0!5" ) --------------- Keys in this file look like this: --------------- LIST BIN.QUEUE 11:49:34 AUG 25 2005 1 BIN.QUEUE..................... 001!001!O!13738!AAL2C!AAL2C 001!001!C!13591!XL2C!XL2C 001!001!C!13591!XM2A!XM2A 001!001!C!13591!XN3C!XN3C 001!001!C!13591!XO3A!XO3A 001!001!C!13591!XP2B!XP2B 001!001!C!13591!XP4C!XP4C --------------- Therefore, given that BTREE_DEFAULT is OCONV(@ID,"G0!5") these two SELECT statements should return the exact same results: --------------- :SELECT BIN.QUEUE WITH BTREE_DEFAULT = "001!001]" 671 records selected to list 0. >CLEARSELECT :SELECT BIN.QUEUE WITH @ID = "001!001]" 638 records selected to list 0. --------------- And as you can see, they return different things. And immediately after this test and a subsequent CLEARSELECT: --------------- Alternate Key Index Details for File BIN.QUEUE Page 1 File.................. BIN.QUEUE Alternate key length.. 35 Node/Block size....... 4K OV blocks............. 1 (0 in use, 0 overflowed) Indices............... 1 (0 D-type) Index updates......... Enabled, No updates pending Index-Name...... F-type K-type Built Empties Dups In-DICT S/M F-no/VF-expr.... BTREE_DEFAULT V Txt Yes Yes Yes Yes M OCONV(@ID,"G0!5" ) --------------- Note the "Enabled, No updates pending". Then following a BUILD.INDEX BIN.QUEUE ALL: --------------- :BUILD.INDEX BIN.QUEUE ALL One "*" represents 1000 records Building "BTREE_DEFAULT" ... 671 record(s) processed. :SELECT BIN.QUEUE WITH @ID = "001!001]" 642 records selected to list 0. >CLEARSELECT :SELECT BIN.QUEUE WITH BTREE_DEFAULT = "001!001]" 642 records selected to list 0. > And though it was corrected via the BUILD.INDEX, it won't last. Here's what's sitting @ AIX: $ ls -l *BIN.QUEUE -rwxrwxrwx 1 root ud 278528 Aug 25 11:55 BIN.QUEUE -rwxrwxrwx 1 root ud 49152 Aug 24 13:01 D_BIN.QUEUE -rwxrwxrwx 1 root ud 110592 Aug 25 11:54 X_BIN.QUEUE $ Though I didn't illustrate it, there was no L_ item prior to the BUILD.INDEX. And unfortunately I can't change the programming to factor out this index due to the way the vendor application has been programmed. Getting this index to be reliable seems to be the only solution. Also there are no triggers on this file. Help? P.S. Are there any logs available in Unidata that might illustrate some faults in the indexing processor? -K ------- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/