Ray There were some problems with UVTSORT and I think they were all addressed from 10.1.7 (not sure which version you are on).
The release notes state: 7295 It the null value was the first character in a record and UVTSORT was enabled, SSELECT statement with a BY clause may have behaved incorrectly or dumped. This problem has been corrected. If you are past that release please flag it in.... Regards JayJay -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Ray Wurlod Sent: 17 May 2006 04:09 To: u2-users@listserver.u2ug.org Subject: Re: [U2] [UV] Upgrading to UV10.1.x - UVTSORT Config parameter This just in from another UniVerse person in my part of the world... Windows cannot multiprocess. Try it and you'll see. I tested a quad xeon, processor 2 ran at 50%, processor 3 at < 10%, and processor 4 was idle. That was 4 phantoms running all sorting. An artificial situation, but nonetheless proving the point. Longhorn will perhaps cure this, for me I'll stick to Linux (work fine with 4000 processors). However incidental, was we turned off UVTSORT last night, and all of the mysterious SIGKILL errors vanished (this was usually killing "uvpset" - which I had replaced with a "do nothing" program). We are worried: how many people are running a p570 which > 50 LPARS. Sometimes "ps" cannot "see" a process that is known to be running. IBM themselves have asked "why are doing this" (using LPARS in production). 5 LPARS using Universe (one with 500 users), others running god-knows-you-name-it: DB2, Oracle and other shit. We were getting SIGKILL every night, usually uvpset, but also "sh", "clear.file" and "spool", last night was clear for the first time since we moved to the p570. ------- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/Ray ------- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
