home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!cs.utexas.edu!swrinde!sdd.hp.com!mips!mips!munnari.oz.au!metro!extro.ucc.su.OZ.AU!rjmartin
- From: rjmartin@extro.ucc.su.OZ.AU (Trilogy Business Systems)
- Newsgroups: comp.databases
- Subject: Re: Sybase and Interactive woes
- Message-ID: <rjmartin.711764896@extro.ucc.su.OZ.AU>
- Date: 22 Jul 92 00:28:16 GMT
- References: <1992Jul16.173500.1075@asc386>
- Sender: news@ucc.su.OZ.AU
- Organization: Sydney University Computing Service, Sydney, NSW, Australia
- Lines: 37
- Nntp-Posting-Host: extro.ucc.su.oz.au
-
- mosmith@asc386 (Matt Smith) writes:
-
- >We are currently using version 3.2 of Interactive and
- >version 4.2 of Sybase. As the subject implies, we seems to be having
- >great difficulty keeping the Sybase server up. If there is anybody
- >out there using this configuration or could suggest a better one, I would
- >really like to hear from you.
-
- We are currently running 4.2 on SCO ODT 1.1 and SCO Unix 3.2., and while we
- have had our fair share of problems, the system is up around 99%+ of the time.
- Our local sybase people are fairly helpful, although when we get really
- hairy problems they are referred back to the U.S.
-
- Something to be really careful about is the number of streams parameters
- configured in the kernal. The first set of release notes were ** WRONG **
- about how many to set up (I think they just thought of some numbers that
- would never be exceeded). You can use the crash command, and then use strstat
- to see how these are going, also make sure you have a good reliable maths
- co-processor (387 compatable), apparently this can cause problems. One last
- thing to look at is the amount of available free ram, and if all else fails,
- try dumping the database, re-install, and start from scratch. We had some
- problems with a demo machine that used unix files as the database devices,
- and occasionally it wouldn't be able to open the log device, thereby marking
- the associated database as being hopelessly corrupt. I then manually changed
- the database status. This happened a few times, and although DBCC came up
- reporting no problems, we finally got to the point where the server would
- crash at the least excuse. I re-installed with BCP'd data and re-loaded the
- triggers etc, and no more problems. We ** think ** that the reason it was
- crashing in the first place was because we were running NFS, or, that
- we had changed NOFILES from 60, which is a bit of a No No in that release of
- SCO ODT
-
- Hope this helps.
-
- John Martin
- Software Manager
- Trilogy Business Systems
-