eth0: eepro100 eth1-4: quad ethernet, tulip driver eth5-8: quad ethernet, tulip driver The installer only adds eth0-eth6 to /etc/conf.modules; eth7-eth8 must be added by hand. (using kickstart + CD) Unfortunately, I can't test this with RHL7 to see if it still happens, but smells like a feature to me. Perhaps interface numbers up to 15 or 16 or so should be supported? :-)
Request seconded, we had that with two Quad-Tulips, too (IIRC), but I forgot to file that in Bugzilla. :-]
Passed to QA to reproduce.
verified this is still a problem w/Red Hat Linux 7 (in my test config, I only got 5 of 7 interfaces configured by anaconda) ...
This works fine for me, both GUI and text. Brock, try again and show me if you make it fail.
Problem nr. 2: Having >16 ethernet interfaces makes the installer (at least kickstart) freeze without error messages. I only have second-hand report on this. I believe this is caused by the fact kernel only by default allows 16 ethXY interfaces and the detection goes like: --- eth15: Digital DS21143 Tulip rev 65 at 0xd00b3400, 00:80:C8:CA:B9:17, IRQ 7. eth15: EEPROM default media type Autosense. eth15: Index #0 - Media MII (#11) described by a 21142 MII PHY (3) block. eth15: MII transceiver #1 config 3100 status 7849 advertising 01e1. : Digital DS21143 Tulip rev 65 at 0xd00b5000, 00:80:C8:CA:B9:18, IRQ 5. : EEPROM default media type Autosense. : Index #0 - Media MII (#11) described by a 21142 MII PHY (3) block. : MII transceiver #1 config 3100 status 7849 advertising 01e1. --- Note that "eth16" is not given a name. The lack of it might confuse the installer.
the install fails with our only > eth5 combination (two cards x 4 tulip nics each for 8 nics) ... an install only recognizes eth0-eth2 (instead of eth0-eth7) ... showed this to ewt, it appears (in our internal case) to be a tulip driver issue ... (at least ewt says so!) ... reassigning to the kernel component ...
tulip driver thinks it deals with up to 8 interfaces Sounds like it isn't working that way for you. I'd like to know what ewt thinks the bug is so that we can look into it better.
The driver (or rather, the ethernet sublayer) should be able to handle up to 16 interfaces. 8 is the limit for passing options=, full_duplex= etc. flags to one kind of driver in /etc/modules.conf.
The tulip driver handles any number of interfaces (even more than 16) just fine. Can we confirm this is still a problem in Red hat 8.0?
I haven't had access to that many interfaces lately, so cannot confirm. Tulip driver does not handle a large number of interfaces in 2.2 kernels; 2.4 is likely better. The first problem may also have been with the installer (6 interfaces added, instead of the maximum).
Thanks for the bug report. However, Red Hat no longer maintains this version of the product. Please upgrade to the latest version and open a new bug if the problem persists. The Fedora Legacy project (http://fedoralegacy.org/) maintains some older releases, and if you believe this bug is interesting to them, please report the problem in the bug tracker at: http://bugzilla.fedora.us/