osvaldo@apollo.la.locus.com (Osvaldo Gold) (08/01/90)
I saw some discussion a little while ago about SCO TCP/IP for Xenix 386 loosing a lot of of packets, being a dog, ftp crashing, etc. Well, that dog just bit me. I just brought up a 386 box running SCO Xenix with TCP/IP installed on an ehernet network with machines mainly running AIX. I get about a 22% packet loss !!!! The Xenix box already panic'ed twice. Does anybody know what is going on??? ---------- Osvaldo Gold - Locus Computing Corporation osvaldo@locus.com uunet!lcc!osvaldo DISCLAIMER: Opinions are only my own.
rogerk@sco.COM (Roger Knopf 5502) (08/02/90)
In article <14215@oolong.la.locus.com> osvaldo@apollo.la.locus.com (Osvaldo Gold) writes: > I just brought up a 386 box running SCO Xenix >with TCP/IP installed on an ehernet network with machines mainly running AIX. >I get about a 22% packet loss !!!! The Xenix box already panic'ed twice. > >Does anybody know what is going on??? Its hard to say for sure without many more details. I know of at least one instance where significant packet loss was traced to a hardware incompatibility between the network card and the transceiver. Best get your hardware and software vendors involved. Problems like this can be tough to track down. -- Roger Knopf <standard disclaimer applies> SCO Consulting Services "The True Believers will...formulate uunet!sco!rogerk or rogerk@sco.com a message that even a monkey could 408-425-7222 (voice) 408-458-4227 (fax) understand." --Jeff Tye
romkey@asylum.SF.CA.US (John Romkey) (08/04/90)
You didn't say what kind of ethernet board you're using. If you're using a 3COM 3C501, expect to lose lots of packets. The 3C501 is a dog. Architecurally, it's the same as the very first ethernet card available for PC's (the 3C500), and has only a single packet buffer, so the receiver must be disabled while loading the buffer for transmit, and you cannot receive back-to-back packets (or anything even close to them). Granted, the TCP shouldn't cause a panic under any circumstances, but for years, th e 3C501 has been a leading cause of poor performance... - john romkey USENET/UUCP: romkey@asylum.sf.ca.us Internet: romkey@ftp.com King Kong died for your sins.
flynn@clotho.acm.rpi.edu (Kevin Lincoln Flynn) (08/08/90)
We had difficulties, where I work, with SCO TCP/IP failing in what may be related. If the SO_LINGER option was not set on a TCP/IP connection, sometimes it worked, most of the time it lost a _lot_ of stuff when the connection was closed. It seems that this bug has been fixed in TCP/IP Runtime 1.0.1h. The other suggestions that have been made don't quite jibe with my experiences with Wyse 386s running 3C501s. However, getting your vendors involved is surely an excellent idea. Disclaimer: The above is not necessarily the official opinion of First Albany Corporation, nor that of the RPI-ACM. ---- Kevin Lincoln Flynn flynn@acm.rpi.edu, userfwvl@mts.rpi.edu 113 9th Street H (518) 272-0048 W (518) 447-8561 Troy, NY 12180 ...Argue for your limitations, and sure enough they're yours.