> I need to create some space here, so it's time for these machines to fit > into the rack with the others. I'm looking to replace the hardware with > some 1U rack servers. I see rack servers available with embedded Intel > PRO/1000 TX NICs, using the 82544 chip. Does anyone have experience running > embedded 82544s (or other similar Intel chips) with LEAF?
The recent Intel Pro100 series NIC's work OK, but I saddly don't have any GBit ethernet equipment to experiment with regarding the newer Pro/1000's...anyone want to volunteer to send me some? :) > According to the Intel site these take the e1000.o driver. I don't see one > of these compiled for Oxygen, will the one in Charles' 2.2.19 kernel work > with Oxygen May 2001 (2.2.19 kernel) or with Oxygen 1.8 (2.2.20 kernel)? I will warn you that the Intel drivers have been getting trickier to compile. I tried to update to the latest Intel drivers with the 2.2.19-3 release of the Dachstein kernel, but was unable to get them to happily cross-compile on my (fairly ancient) Debian development system. This probably has more to do with not having the latest version of some of the building tools (I had to update make to get even the older Intel drivers to work) than with Intel's code, however. Also, I think Donald Becker has drivers for the GBit Intel cards, and his drivers are almost always easy to compile and usually work quite well... > Also, these are SMP-capable boxes. I won't order them with two processors > for this application, but am I going to run into problems using a > uniprocessor LEAF kernel on SMP-capable machines? There shouldn't be any issues. In fact, AFAIK, most PC's sold today would fall into the catagory of "SMP capable", since support for 2-way (in some cases 4-way) SMP is essentially built into the processors (as well as most chip sets). To turn a conventional PC into a 2-way SMP machine, you basically just have to re-layout the board (adding another CPU socket), be *REALLY* careful about your trace lengths/impedence/delays, and probably strap a few lines differently. If you're not talking about custom chipsets for 8 way SMP or something, there shouldn't be a problem (famous last words :) > Some other machines I've looked at are single-processor designs with > embedded Intel PRO/100+ NICs (82559 chip). I guess these would be either > eepro100.o or e100.o drivers. Are these machines a better choice? > Obviously there's more room for growth in the others, and the prices are not > that different. Hmm...if the prices arn't that much different, I'd probably go with the spiffier systems. Especially if they use a 'normal' desktop chipset you can verify LEAF (or linux 2.2.19) runs on successfully. Worst case, you might have to compile a custom kernel (or get someone to do it for you). > I know this is different harwdware from the majority of LEAF installations, > so if anyone knows about any other "gotchas" in LEAF on a rack server, I'd > be grateful. Wear your gloves! Whenever I'm messing with rack-mount stuff, it seems like I always manage to slice up my hands, even though there arn't any sharp edges on my (re-purposed 1/2" mainframe tape drive) DEC rack... > Sorry for the dumb questions. Not dumb questions, and it's always fun to hear about new purposes/platforms LEAF has made it's way to. Charles Steinkuehler http://lrp.steinkuehler.net http://c0wz.steinkuehler.net (lrp.c0wz.com mirror) _______________________________________________ Leaf-user mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/leaf-user
