Kit Haskins shares some of his recent (May of 1999) experiences with Red Hat Linux.
Within the past two months, I've set up a Linux Redhat 5.2 box on my local LAN to see what all the hype was about. It was a learning experience, considering most of my *poor* unix skills are based on SCO ODT and Openserver, from Jeff L's "learn by destroying" methods here at Basement University.
The network has a pair of SCO Openservers, one of them is a gateway for the PPP connection to my ISP and the second SCO Openserver has the HP 6P printer attached and runs LPD nicely. The two SCO Openservers were quite happy until the third unix host, a Linux Box, came to town.
The LPR configuration on the Linux box was a pain. "linuxconf" would ask for the printer name, and hostname and do its magic. Having no luck in getting paper to come out of the printer, I checked the SCO Openserver for /etc/hosts.allow and /etc/hosts.lpd for anything goofy. They still liked each other, printed requests from each other, the entries for the new addition seemed to be correct, it just ignored the penguin.
The final fix was the /etc/printcap file on the Linux box had the rm and rp fields swapped. Meaning the printername and hostname in the printcap were in the wrong places. Now the Linux box could print via the SCO Openserver, text only but it was a start. Having that repaired I got a laugh provided by the warning in the header of the file:
"Please don't edit this file directly unless you know what you are doing! Be warned that the control-panel printtool requires a strict format!"
Now that the printer problem was repaired I would continue on with X-Windows. I wanted it acting as a X-Terminal and religiously following the documentation that seems to be scattered all around the hard drive like configuration files for a Windows 95 system.
The Linux box would run just find and dandy to only one of the SCO Openservers, the one with a i486 processor. Starting an X session on the Openserver powered by a Pentium II would only get as far as blanking the screen prior to the login prompt. Having two near identical hosts, a file comparision and check of read/write permissions yielded nothing different. It is nice to have something to compare with, even if it only kills 7 hours looking.
"netstat -p tcp" showed the Send-Q of the Pentium Openserver had 14600 bytes in queue. Figuring it was a network problem I starting to pull extra devices off the LAN to minimize traffic. Didn't help. Reconfigured the card for different IRQ's and Base IO address, I swapped out the 3COM 509 card with an identical 3COM 509 card, nothing helped. FTP transfers also seemed to be slower on the Linux box, with 400Kbps thruput on a the LAN compared to 4Mbps with the Openservers to each other.
I figured as one last ditch effort to replace the NIC in the Linux box with an old Lantastic card jumpered to emulate NE2000. The next two hours was spent in getting the driver module replaced with the correct version. Linux doesn't seem to like change that much. The older NIC not only sped up the FTP transfer rates but also acknowledged all the packets in the Send-Q and ran X sessions without any further hassles.
So in my recent "learn about the hype" adventures I've discovered:
I'm not going to look forward to getting eFax working.
© May 1999 Kit Haskins
Got something to add? Send me email.
More Articles by Kit Haskins © 2011-05-01 Kit Haskins