erickson@pollux.ncgia.ucsb.edu (Scott Erickson) (08/12/90)
proot@ksuvax1.cis.ksu.edu (Paul T. Root) writes: >I basically like the machine [RS6000] a lot. But the little things >like [sendmail] and the ypbind/inetd problem really leaves a bad taste >in your mouth. I find that (on IBM RT's, PS/2's, and RS6000's) IBM tends to make their software so that it's easy to connect with other IBM machines, but it's extremely painful to connect to anything else! YP (NIS), sendmail, NFS, etc. seem to work ok between our two RT's, but it took a major struggle and lots of options to get this stuff to work between the RT's and Suns. And simple programs like "install" don't work right on the RS6000. All of my favorite switches like -o, -g, -m have been changed to upper case. Now, I'm sure these some mode or path somewhere you can set to get the real install, but if you login to one of these systems with simple program source and type "make install", it won't! Sure, each OS has its peculiarities. I just find that AIX 3 seems to be too different in some of the *basic* programs (like install). I've had nothing but bad luck with the pre-releases OS's crashing the system, and now that at a fairly stable OS is out, I can't just bring my local software over without lots of pains. If I was still a sysadmin with a Computer Science Department, I'm sure I'd enjoy all of the neat new things like the access control lists and the hypercard libraries and such, and my users would probably be just as interested. But being with the Geography Department, I'm just interested in getting a system up so that my users can get some work done. My users just want a machine to help them with research. They don't want to have to deal with all of these problems on a daily basis. I've put on my Big Blue Asbestos underwear. If you're going to flame, please just send mail to me. I'll "summarize" to the net ;-) /se University of California Santa Barbara | (805) 961-3712 Department of Geography | erickson@pollux.ncgia.ucsb.edu Support Group | The opinions expressed here do not necessarily reflect those of any entity either living or dead and most certainly do not necessarily reflect those of the University of California or its Regents. -- Scott Erickson
jeffe@sandino.austin.ibm.com (Peter Jeffe 512.823.4091) (08/13/90)
In article <6101@hub.ucsb.edu> erickson@pollux.ncgia.ucsb.edu (Scott Erickson) writes: >I find that (on IBM RT's, PS/2's, and RS6000's) IBM tends to make >their software so that it's easy to connect with other IBM machines, >but it's extremely painful to connect to anything else! > >YP (NIS), sendmail, NFS, etc. seem to work ok between our two RT's, >but it took a major struggle and lots of options to get this stuff to >work between the RT's and Suns. Hmmmm. I can't speak for YP/NFS, but in the tcp/ip group we've had few problems with interconnection. Judging from the sorts of problems reported in the tcp/ip and sendmail newsgroups, AIX seems to have less than its share of inter-vendor connectivity problems in these areas. Since the majority of our tcp/ip and mail code is straight BSD tahoe ports, I don't find this too surprising. I would, however, be interested to hear precisely what sorts of problems you've had. Please respond by email unless you consider them of global interest. ------------------------------------------------------------------------------- Peter Jeffe ...uunet!cs.utexas.edu!ibmaus!auschs!sandino.austin.ibm.com!jeffe first they want a disclaimer, then they make you pee in a jar, then they come for you in the night
steve@qe2.paloalto.ibm.com (Steve DeJarnett) (08/14/90)
In article <6101@hub.ucsb.edu> erickson@pollux.ncgia.ucsb.edu (Scott Erickson) writes: >proot@ksuvax1.cis.ksu.edu (Paul T. Root) writes: >>I basically like the machine [RS6000] a lot. But the little things >>like [sendmail] and the ypbind/inetd problem really leaves a bad taste >>in your mouth. sendmail.cf files seldom, if ever, drop out of the box and work. IBM tried to make it so that you wouldn't have to change much, but anyone who has anything other than a generic setup shouldn't be too surprised if they have to install a new sendmail.cf file. >And simple programs like "install" don't work right on the RS6000. >All of my favorite switches like -o, -g, -m have been changed to >upper case. Now, I'm sure these some mode or path somewhere you can >set to get the real install, but if you login to one of these systems >with simple program source and type "make install", it won't! Well, /usr/ucb/install has everything you're asking for. As a rule, if you used BSD or BSD-derived systems before, and that's what you want from your RISC System/6000 (RS/6000), then ALWAYS put /usr/ucb first in your path. That's what all of us "BSD-bigots" here do. :-) >Sure, each OS has its peculiarities. I just find that AIX 3 seems to >be too different in some of the *basic* programs (like install). >I've had nothing but bad luck with the pre-releases OS's crashing the >system, and now that at a fairly stable OS is out, I can't just bring >my local software over without lots of pains. Again, most of the major programs that have incompatibilities have BSD versions in /usr/ucb that should behave just the way you (and I) are used to. >But being with the Geography Department, I'm just interested in getting >a system up so that my users can get some work done. My users just >want a machine to help them with research. They don't want to have to >deal with all of these problems on a daily basis. Put /usr/ucb in the first position on their paths and they won't have to worry about it. I'm not saying this will solve 100% of your problems, but it should go a long way towards it. >Scott Erickson Steve DeJarnett Internet: ibmsupt!steve@uunet.uu.net IBM AWD Palo Alto UUCP: uunet!ibmsupt!steve (415) 855-3510 VNET: dejarnet at ausvmq These opinions are my own. I doubt IBM wants them.......
karish@mindcrf.UUCP (Chuck Karish) (08/15/90)
In article <6101@hub.ucsb.edu> erickson@pollux.ncgia.ucsb.edu (Scott Erickson) writes: >And simple programs like "install" don't work right on the RS6000. >All of my favorite switches like -o, -g, -m have been changed to >upper case. Now, I'm sure these some mode or path somewhere you can >set to get the real install, but if you login to one of these systems ^^^^ >with simple program source and type "make install", it won't! All the world's not a VAX (though AIX executables do have zeros at address 0), and there's no one, true UNIX that all good implementations must slavishly follow. The changes were made to avert conflicts with flags used by System V versions of install. If they hadn't been made, some System V users would be complaining just as loudly. >Sure, each OS has its peculiarities. I just find that AIX 3 seems to >be too different in some of the *basic* programs (like install). install is a basic program? I usually use cp or mv instead; they're more portable, and install is inexcusably slow on some systems. >I've had nothing but bad luck with the pre-releases OS's crashing the >system, and now that at a fairly stable OS is out, I can't just bring >my local software over without lots of pains. I don't recall whether they've been mentioned here before, but IBM has made available two guides to AIX 3.1 for 4.3 BSD users. They're called `AIXv3 for 4.3 BSD System Administrators' and `Porting 4.3 BSD Programs to AIX Version 3.1'. They're a bit big (100K and 65K, respectively), but if there's enough interest it might be worthwhile to split them up and post them. -- Chuck Karish karish@mindcraft.com Mindcraft, Inc. (415) 323-9000
shair@ux1.cso.uiuc.edu (Bob Shair) (08/16/90)
karish@mindcrf.UUCP (Chuck Karish) writes: > I don't recall whether they've been mentioned here before, but IBM > has made available two guides to AIX 3.1 for 4.3 BSD users. > They're called `AIXv3 for 4.3 BSD System Administrators' and > `Porting 4.3 BSD Programs to AIX Version 3.1'. They're a bit big > (100K and 65K, respectively), but if there's enough interest it > might be worthwhile to split them up and post them. >-- Those with Internet access can use anonymous ftp to obtain these documents from vmd.cso.uiuc.edu (128.174.5.98) cd to directory SHAIR.3CC and get the files bsdportx.tr and bsdadmx.tr. They're text documents which are to be processed by troff to print, use "me" macros. Bob -- Disclaimer: I am not speaking officially for IBM or the University of Illinois in this append Bob Shair shair@vmd.cso.uiuc.edu Scientific Computing Specialist SHAIR@UIUCVMD (bitnet)
madd@world.std.com (jim frost) (08/23/90)
In article <6101@hub.ucsb.edu> erickson@pollux.ncgia.ucsb.edu (Scott Erickson) writes: >I find that (on IBM RT's, PS/2's, and RS6000's) IBM tends to make >their software so that it's easy to connect with other IBM machines, >but it's extremely painful to connect to anything else! We ran into a neat one yesterday. If your root shell is /bin/ksh (the default) and a user's home directory is NFS mounted (most of our users are on a Sun-4/330 under 4.0.3c) and the user "su"'s to root, ksh will start making tons of NFS requests across the network. It will flood the network, thus inhibiting everyone's communication, and bring the foreign server to its knees (biod's will be running all-out). Killing the ksh process started by "su" causes the problem to go away. Changing the root shell to /bin/csh eliminates it entirely. This problem exists under 9021F with the 3001 update installed. We never noticed it before the 3001 update so I believe the update caused the problem (might be wrong though). This problem has been reported to IBM. Informationally yours, jim frost saber software jimf@saber.com