Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!rutgers!bpa!cbmvax!vu-vlsi!dsinc!lgnp1!vskahan From: vskahan@lgnp1.LS.COM (Vince Skahan) Newsgroups: comp.sys.apollo Subject: What's going on at Chelmsford (...long...) ??? Keywords: which inmate is in charge of the asylum ??? Message-ID: <986@lgnp1.LS.COM> Date: 6 Apr 89 03:07:55 GMT Organization: Boeing Computer Services - Phila. Lines: 165 [... frustrated flame on...I welcome comment but please lets not start a total flame war...I'll be happy to summarize any responses if there are lots of them...I just want to voice my concern to the folks in Chelmsford and I realize that most of the APollo folks who post to USENET are concerned, experienced, and able...] Can someone in Chelmsford please explain a few things to me ??? Over the last couple of years I've complained a little about Apollo's (ssemingly) silly quirks but I've almost always defended them to the death over anything else. I've seen nothing better but things can be improved. I've dealt with high turnover and low expertise on the hot-line as people came on board. I've put up with the stinkin' vt100 emulator that in my observation continues to get worse, not better. I've raved about how wonderful RAI is and how easy it makes my job. I've screamed because some of the SR10 versions (DPCC 3.0 for one) are not available in RAI format, making the use of an authorized area impossible (even though all SR10 versions are SUPPOSED to be in RAI format so we can hard-link). I've asked when RAI setup instructions are going to be made available to the general public to make third-party and local software installations just as easy and am living reasonably with the answer "it will be coming someday". I'd like to know which marketing genius is getting promoted in Chelsford for the following recent (dumb in my personal view) changes in focus and perpetuating the silliness of the past: - changing the DN3000 to have only one memory slot that takes only 4 or 8MB boards. Sure...they'd like us to all buy 4500's with 32 MB of memory and 700 MB disks but some of us have dozens of old 3000's and we need seats, not mips. What am I supposed to do with the old 4MB boards when SR1x.x comes up that takes more than 4MB of memory??? I'll be out of slots, out of capital, and floating in $2900.00 coasters for my Diet Coke cans. - how am I supposed to run anything other than the OS on a DSP90 that is maxed out at 3MB of memory. I have little confidence that I can run TCP, the OS, and a print server on 3MB. God help me if I try to run something that's memory intensive like a Cadre Database-controller server. What are the poor folks with DSP80's that are limited even more supposed to do ??? I could tolerate buying 4 generations of ethernet boards a couple of years ago within 9 months (com-eth, ecmb, eth-at, a-net-eth) but I can't handle planned/unplanned obsolecence of tens of thousands of dollars worth of hardware. I think within the next year or so I'll have to remove the hardware, drill a hole in the top, and make it a $25000 beer-meister. - shipping SR9.7 X-windows in RAI format not in the (old) install format. The release notes state that a full server installation takes 37 MB for the AA. Since you have to install from the AA (not tape) and you can't hard-link under SR9, it states that it's another 37 MB for the real in-place software. Sure...it says you can then delete the AA but where am I going to come up with 74 MB of disk space (plus the temporary room that RAI seems to need) and once I delete the AA to save room, what do I do when 2 months later someone else decides their project needs X too. - shipping SR10.x in my belief over 9 months before it was stable. I just got a 30 or so page set of notes from Apollo identifying bugs, misleading instructions and questions, and non-documented workarounds for SR10 installs only. It's ridiculous to me that these problems exist (although I DO give APollo some credit for being proactive and telling people (late is better than never))... you should do it once, do it right, and not jerk everyone around promoting real unix when you aren't ready yet...and the terrible thing is it'll happen again wehen the OSF gets their act together. - Why oh why does Apollo continue to refuse to release known bugs to the general public and expect us to have something fail, pull our hair out working with (and training ) the hotline before they kick it to R+D and 2 weeks later agree "yep...it's broke..." ??? Mentor Graphics sends out monthly sys_admin bulletins with identified problems and workarounds/statuses for many known bugs. I've seen 95 % of them already by the time the mail gets to me, but it DOES explain why it's broken and when it'll be fixed for their 3rd party software. Hey Apollo, make it look at least like you're trying a little, OK ??? - although documentation is getting better, why is it all still so lousy (JLRU, eh?). I'm still waiting to see in one book how to set up a ring correctly rather than learning it the hard way. I've been with Apollos since SR9.2.3 so I can do a lot by sheer experience but how do I train my replacement when he wants to see something clear, concise, and complete in writing and it doesn't exist unless I write it myself ?? if you're going to go to just like real unix, why did you ignore the details of basic but important things??? I agree completely with supplying third party books (like with the X product) but the Apollo-specific stuff is incomplete. Why do you devote about 3 pages to lpr when you should devote 23 ??? Why do you totally ignore things like printcap, termcap and apollo-apollo ring configuration. I've got 11 rings to deal with...why don;t the manuals say something about the timedelays in having 11 local site registries...it takes over 90 seconds sometimes to get the edacct> prompt back while opening all those sites. Where is the discussion of how to set up unix mail in a multi-ring internet and why are the sample sendmail.cf files so awful ??? The above are only a few unix-related examples since as one co-worker observed, unix is passed on from guru to apprentice by oral tradition only. The same (and more) apply for Aegis. - why did you make such a big deal about lprotect and then send the next OS out with "don't use this or else..." in big letters ??? If it's not ready, don't relase it until it is. - why did you change the node ACLing procedure when the old one wasn't broken and then give absolutely no example templates to have us run against the existing software to change the level of protection ??? Telling me to re-invol is just not good enough. That's bad enough but then I have to find out that in SR10.1 "open" means closed and "closed" means open ??? Please... (yet another instance where I heard it first and best on USENET...thanks to the thoughtful person out there whose name I misplaced who mentioned it here ). - why do I have to find out that patch tapes and PSK's are available third hand ??? I run the OS, I pay maint, I should get the tapes without asking and without searching. I'll take responsibility for deciding whether or not to install a particular patch. On that subject, why doesn't Apollo send out mandatory patches ??? Some of us are bound by third party software and can't immediately install a major release for a while...my SR9.6 sat in the box for 8 months...I have 5 of 80 nodes at SR10 now (9 months after its release) and won't upgrade everything until about November or so at the earliest. What am I supposed to do for my rings that are on segregated projects running one version of the software who will be at 9.7 for 18 months or so??? - who came up with the idea for spiral-binding the manuals ??? My hard-negotiated "right-to-copy" the manuals isn't worth the paper it was written on. - Since I've seen that RAI patches get deinstalled when you re-install them (real smart guys...), why can't I determine exactly what is installed and why can't you provide a de-install option for packages ??? Ultrix has a lovely feature that identifies what packages are installed and permits de-installs. - why does Apollo continually say that their disks don't fragment. Try using a DN3000 in a diskspace-bound ring for 18 months. Then invol one node and reinstall the same products. Then tell me that there is no difference. Why can't I tell how fragmented my disks are (I know...because they're not, right ???)? [...flame off finally...] I'm sick and tired of beating my head against a wall...I'd appreciate hearing some response from some APollo folks (and any comments from anyone else) regarding what the heck's going on and why ??? I don't want to buy from other vendors but Chelmsford's going to have to do a better job in having some more consideration for their installed customer base. You can reach me at (215) 591-4116 during working hours or by e-mail at bcsaic!vince or vince@atc.boeing.com Like I said, I'll summarize any responses I can if there are enough that haven't been said before... Thanks for bearing with me... -- Vince Skahan - please reply to skahan@boeing.com or bcsaic!psev!bcs212 Note: any comments expressed above are mine and have no relation to Boeing or the real nice folks who let me read news on their system...