home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.unix.xenix.sco
- Path: sparky!uunet!uunet.ca!geac!torag!zooid!vid
- From: David Mason <vid@zooid.guild.org>
- Subject: Re: Still having missing character problems
- Organization: ZOOiD BBS
- Date: Mon, 16 Nov 1992 15:02:02 GMT
- Message-ID: <1992Nov16.150202.12019@zooid.guild.org>
- References: <1992Nov12.144923.725@zooid.guild.org> <Bxo9vo.Kqq@world.std.com>
- Lines: 45
-
- ahmcs@world.std.com (Alan H Mintz) writes:
- >David Mason <vid@zooid.guild.org> writes:
- >
- >>A while ago we had some problems on our SCO 2.3.2GT (upgraded with the SCSI
- >>enhancement) system, running on a 386-33, 64k cache, 8MB RAM, Adaptec 1542B,
- >>IDE & SCSI disk, AST 4port, standard I/O, FAS 2.09, monochrome system (whew).
- >>The problem is that we are getting characters disappearing from input and
- >>output on both the serial lines and the consoles. This will usually happen
- >>after about a week or more of 24 hour uptime. Rebooting the system once a week
- >>is a solution, though not a very good one.
- >
- >>When I originally posted this question, several people suggested upping the
- >>number of CLISTS on the system. Originally it was set to the default (100 or
- >>200). We have 5 serial ports plus the consoles so 300 seemed like a
- >>reasonable number, but the problem persisted. I upped the number to 400 but
- >>characters are still disappearing after about a week of uptime.
- >
- >I don't know if I responded previously or not, but I don't think your
- >problem is CLISTs. When we went from a 386/25+Adaptec 1542 to a 486/50+
- >Dell DDA (EISA SCSI-looking thing), we started seeing the same problem under
- >2.3.4.
- >
- >It seems to coincide with heavy disk load, leading me to believe that
- >the SCSI controller and/or driver is "traffic-jamming" the bus and/or kernel
- >to where serial interrupts are being lost. Does this seem reasonable?
- >
- >We only see it under VERY heavy loads, so I haven't pursued it. Still, a
- >solution would be nice.
- >--
-
- It's possible, but the problem doesn't go away after heavy disk useage. It's
- gone on for several hours while the system has been reasonable idle, until we
- rebooted. Some resource is being grabbed during heavy SCSI useage, and not
- released afterwards.
-
- It must be a Xenix bug, I've never heard of this happening on any other
- platform, and Adaptec 1542B is about as standard as it gets...
-
- I think the fact that it happens after about a week of uptime is a clue, but
- otherwise I'm clueless.
-
- Any other ideas?
-
- --
- vid@zooid.guild.org -->> ZOOiD Public Access Usenet 416 322-7876
-