The collected works of josehill - Page 1

Hakimoto wrote:
unixmuseum, great stuff. can you post the relevant policy.conf sections by any chance? Would be good to see if this stuff is commented or not...

I'm not sure if this is what you are looking for, but the man pages for several Solaris (back to version 2.4!) are searchable at , and you can browse to the man page for policy.conf in the Solaris 10 "Reference Manual" section.
If I remember correctly, only the Windows, Solaris, and Linux versions of Oracle 8 were available for download...For IRIX, only the documentation was downloadable...

BTW, the final versions of Oracle for IRIX were 8.1.7 for 64 bit systems and 8.0.6 for 32 bit systems.
CDG wrote: In fact, the more reading you do about the Indy, the more you realize that they weren't used by much of anybody.

I dunno. When the 180MHz R5K WebForce Indys came out, we used one as the "authoring" system for a corporate intranet served from a 180 MHz R5k WebForce Challenge S (the latter was also serving as a router). For 1996, it was a terrific setup, and the combination served thousands of users. CosmoCode, Photoshop, and Illustrator on an Indy combined with a very good network stack and disk I/O on the Challenge (basically an Indy with a second, multi-channel SCSI card and, in our case, a FDDI card) were pretty neat back then.

That Indy worked great, and it's still in service as a terminal running 6.5.22 on a developer's desktop. It's kind of a fun story actually -- the developer codes for Windows and Solaris, and he wanted a second machine for his Solaris work, but he didn't like either CDE or Gnome. I offered him the Indy, and he decided to try it, even though he had no IRIX experience. Now he mounts his Solaris directories on his Indy and does all his coding in the IRIX environment, and he only logs in to the Solaris server to kick off compilations. By the way, he just bought an Octane for home use. The contagion spreads! :lol:

PS. Unfortunately, the CPU module died on the old Challenge S late last year, so that box is officially in "spare parts" mode...
Okay everybody, you can breathe a sigh of relief.

It seems that is back online as of ~ 9 AM California time.

Maybe they laid off one too many sys admins? :wink:
ipaddict wrote:
jfeldt wrote: I have a Linksys WAP54G that I have been using as an access point client. It has been totally stable for as long as I have owned it (~8 months). Whatever you end up getting, make sure it can behave as a client also, since I think that isn't a given, as one would believe.


That is referred to as a "bridge." :wink:

True enough, but a lot of the wireless manufacturers insist on calling it "access point client mode"or something similar. When talking to tech support, they often won't have a clue until you use the magic sentence: "I want my access point to operate in client mode."
dir_marillion wrote: Hello,
I am wondering if there is a remote desktop software from Windows XP side that could control Irix O.S. or the opposite, to a local network.

If you are running 5.3, it might be a challenge to find a vncserver package that will work. For 5.3, it will be much easier to run an X Windows server on your PC.

There are a few free X Windows packages that will run on XP. You might want to try X-Deep/32 . Another good option would be Cygwin/X , but it might be a little more complex to set up. You might also take a look at MI/X which is simple, free to try, and $25 to buy.

VNC does have certain advantages, however, and it is easy to set up if you are running 6.2 or newer. You can download an older version of vnc for 6.2 from sgi.

You can get nice 6.5 versions of TightVNC either from nekoware or from the 6.5 repository at . If the latter, be sure to read the release notes.

If it is 5.3, you might have more of a challenge. I'm not sure how easily you will find the vnc software -- you might need to get the source code and try to compile it (or convince someone else to try). :(

If you choose the VNC approach, I have had good luck with TightVNC running on XP.

Finally, as Joerg suggested, you can always just open a shell window on XP and connect via telnet (or use HyperTerminal via a serial connection).

Good luck!
dir_marillion wrote: josehill, do you know something about the existence of "ported" tightvnc for mips3 ?

I think that the IRIX 6.2 version of vnc at SGI Freeware is mips3, but I'm not 100% certain, and I don't know if it will run under IRIX 5.3.

Sorry I don't have any other info.
Thomas W. wrote: :cry: :cry: :cry:

That was it for IRIX?

That was "it" for manufacturing current IRIX hardware , though it's a pretty safe bet that there won't ever be any new IRIX hardware, either.

As for IRIX itself, the announcement goes on to state that
The earliest projected end-of-support date for new Origin 350 and 3900, Fuel and Tezro is currently December 2013. there will probably be at least one lonely IRIX engineer left until then.
yetanother**ixuser wrote: hmmm....not so bad, since my learning curve will drop cause only solaris, aix and of course linux are left for production use.
Hey, there's still HP-UX! :wink:
psergiu wrote:
josehill wrote:
yetanother**ixuser wrote: hmmm....not so bad, since my learning curve will drop cause only solaris, aix and of course linux are left for production use.
Hey, there's still HP-UX! :wink:

HP/UX on Titanicum is a sorry hack and the PA-RISC cpus are killed as we speak. :x

Okay, okay! I take it back! I never said anything! :lol: :lol: :lol:
Oskar45 wrote: Ok, 20 years of SGI producing workstations are history, but nevertheless I seem to have noticed that one or the other on Nekochan still happily hacks away on boxes for which the end-of-support date has long passed but none of them really complains much about lacking support - so why lamenting now? I myself am reasonably sure I'll get support for my Fuel for the next 10 years or so as well [since my local SGI support still maintains boxes from 1994, I'm confident I'll manage a support contract beyond 2013 as well].

I tend to agree with you, but it does mean that Nekochan will become even more important to the community than it is now! :D

Oskar45 wrote: And - I already have this crazy idea to get me a remarketed Tezro next year - I've no idea what do to with it yet, but it would sure be cool to own one :D

Yeah - me too!

Oskar45 wrote: Besides - noone of us can really tell what the *nix landscape will look like in, say, 10 years from now - perhaps then there will be none anyway and everything is just a nightmare, eh Windows :)

Nah, we'll have direct neural implants running embedded BorgOS by then. :lol:
IRIX 6.5.x has been getting regular upgrades since 1998 -- it's a very mature platform.

Of course, it has some serious gaps in a few areas like USB, Firewire, the desktop GUI, etc. Even though those areas are important to hobbyists, they are generally not high priorities for SGI's big $$ customers. SGI is struggling, and filesystem tweaks, I/O tweaks, etc., are what the customers with million dollar service contracts are asking for, and that's where SGI's diminished IRIX efforts are sure to be focused.

For those who aren't interested in such things, it's probably best to start looking at new IRIX releases as consolidated patch sets, rather than as major releases .

All things considered, that's still pretty decent support from a company going through bankruptcy for an OS that's probably not too far from going into maintenance mode (i.e. security/bug fixes only).
foetz wrote:
josehill wrote: Of course, it has some serious gaps in a few areas like USB

who cares? oh and there are usb keyboards and mice :D

I do, a little bit. :wink:

It would be nice to be able to buy a common USB2 card, slap it into any PCI SGI box (like an O2, or even an O2K, not just the more recent machines), and then connect one or two $250 500 GB USB2 drives and have them work without having to sacrifice a goat first, rather than pay the same amount for a 73 GB SCA drive.

I prefer using SCSI drives for servers and critical applications, but for certain uses, it would be great to have a reasonable, cheap USB storage option, especially on older hardware.

foetz wrote:
josehill wrote: , Firewire


Sure, but poking around in other Nekochan threads reveals caveats about one chipset or another, things that work on Tezro or Fuel but are more, um, complex on other systems, and so on. No doubt it's great for some, but Firewire is one of those things that should "just work."

foetz wrote:
josehill wrote: , the desktop GUI

what's wrong with the dekstop? it's perfect! productivity oriented, very fast, slim, not overloaded, stable ... it's just not a toy like desktop today. but exactly this is one of the major pros at least for me.

I agree! (Well, maybe I'd only say it's "really, really great!", not "perfect!" :lol: )

I'm not talking about adding all sorts of wizzy effects or bubblegum and eye candy, but really just some tiny, gradual improvements in functionality. To be clear, I am also referring to the entire desktop application environment, not just 4dwm. In many ways, KDE and Gnome still have not reached the level of functionality and behavioral consistency that IRIX 6.5 had in 1998. Unfortunately, as IRIX desktop development efforts declined, some of the beautifully integrated functions have been replaced by less elegant (but perhaps more complete) open source solutions (for example, Impressario vs CUPS). If IRIX desktop development had continued, I could imagine all sorts of refinements for dragging-and-dropping between desktop applications, as well as many other things that might be useful. A moot point, however, since the desktop certainly isn't getting any more refinements!

By the way, I really wasn't complaining about IRIX. The quality of the IRIX Interactive Desktop is the main reason why my staff and I continue to use IRIX desktops as our preferred frontend to our Linux and Solaris systems!
nekonoko wrote:
josehill wrote: A moot point, however, since the desktop certainly isn't getting any more refinements!

Hmm, I disagree with that point. From SGI itself, yes, but I'm continuously amazed at some of the refinements that the SGI user community has come up with - iconbar, imwheel and aer (Advanced Enforce Rate) to name a few.

Of course, you're right about that. I was only referring to official SGI products.
foetz wrote:
sure but that's not what the typical sgi customer cares about and i'm happy about this.

Actually, that was a central point of my post...SGI has to focus on what its biggest markets want, and certain common technologies aren't at the top of that list, so SGI is doing exactly what it should be doing. :wink:

...but it's also clear that the absence of full support for such common standards as USB2 and Firewire is a visible reminder of SGI's loss of competitiveness in the general Unix computing market (as opposed to certain niche markets) compared to Solaris, Linux, or even HP-UX.

At least AIX is still pretty lousy with USB and Firewire, too... :lol:
foetz wrote:
okay but where is usb2 a common standard in hpc?

I'm not arguing that USB is a requirement for hpc, just that it can be a real convenience . It is showing up more and more in the marketplace, and it adds some flexibility to an environment to be able to just plug in a USB device when necessary.

Certainly, if you are buying Opteron or Intel based HPC nodes, they will have very well supported USB ports on them, whether you are running Linux or Solaris, and it's kind of neat to think about booting from a thumb drive for quick troubleshooting. It also seems to be well supported on Sparc systems, but even if your device isn't supported, just go to the Sun website and get the Solaris USB Driver Development Kit and write your own device driver! While it's not trivial to implement, full USB support isn't exactly rocket science, either. (I'm joking a little there, but the easy access to such tools shouldn't be trivialized. :D )

Interestingly enough, for some of my common hpc tasks, I receive large data sets from my collaborators on 500 GB USB drives -- it's much more convenient to just plug such a drive into an hpc node than to shuffle through 10 DVDs, or even to connect the drive to a workstation and then export to an hpc system via nfs. Sure, you can do those things without a lot of trouble, but why jump through hoops if you don't have to? When plug-and-play works, it's a great thing!
foetz wrote:
anyway i never said it wouldn't be handy. just not prio 1.

Agreed! As always, best wishes... :D
zizban wrote:
Ever try just plugging in a usb flash drive into a solaris box? it doesn't just work. You do have to sacrifice some small furry animals to get it to work.

When you tried, was it on Solaris x86 or Sparc? Solaris 10?
...and don't forget to install "Patch 7110 : XVM Failover initialization problem", which SGI released simultaneously with 6.5.30...

Critical Patch for IRIX 6.5.30 XVM Failover

Patch 7110 released with IRIX 6.5.30 to resolve a condition that produces extraneous XVM failure messages at startup. The problem effects failover and setting up the correct paths to targets. There is no data integrity problem. However, this problem could affect performance after a boot. The failover file can be used to balance the load across HBAs, RAID controllers, etc. Because of the problem, it is likely that lots of paths will use the same HBA instead of being spread over all the HBAs. Read/write performance could be impacted, especially on systems with lots of paths to each target. Patch 7110 will be preinstalled on new systems shipping from SGI worldwide manufacturing facilities. Patch installation is also recommended for customer upgrades.
87Porsche wrote: And now they've been pulled from the site as of noon here.

Looks like they are back. Dare we to risk another attempt? :?
fu wrote: yay!

that did the trick

So, are you going to show us the image? :wink:
While there are some docs on the net about securing IRIX and configuring ipfilters, I'd go with OpenBSD for a firewall -- clear track record, excellent documentation, secure "by default", and so on. I'm not saying that it's impossible to use IRIX as a firewall system, just that it will take a lot more effort and knowledge to secure it and to keep it secure.
I only recall seeing the base dwb packages on their own, separate cds, too. It was a separate, commercial product.
SAQ wrote: Groff it is, then. I just thought I might have seen it on that page in the devel. progam with all of the downloadable tools that's disappeared, but I guess not.

I'd heard that it was depreciated with the release of IRIX 6.5 anyway. What does DWB have other than nroff?
Yeah, but they still have DWB overlays on the standard IRIX distributions (just no copy of the base dwb.sw.dwb -- that's what the DWB product cd gives you).

As for what else is in DWB, it includes the usual AT&T nroff, troff, pic, a couple of fonts, and some shell scripts like psroff, IIRC.

Groff should be able to handle most requirements, though.
pipeline wrote:
badapple wrote:
So, can any XFS experts point me in the right direction? What values should I be setting the physical raid stripe/block and XFS filesystem variables to? Should I align the values on the hardware and filesystem exactly - I'm more used to ext2/3 filesystems and '-R stride= x '.

The defaults are the defaults for a reason. I'd just leave it the heck alone.

pipeline's probably right on this one. If you're already happy with performance, then you're probably already done. You didn't mention how the RAID will be used (consecutive reads or writes vs random; large files vs small files; databases, etc.), so the real answer is, "It depends."

If it's really critical to you, no amount of advice beats benchmarking under the type of load you anticipate experiencing...
"gated" has some more flexible capabilities than "routed," but either is fine.

If you want persistent static routes, put them in /etc/config/static-route.options for 6.2+.
What does the output of netstat -rn look like on the Tezro and on one of the machines that can't get to the public net?
thermionic wrote: FreeBSD rshd is now working with inst :-) It needed several changes to /etc/pam.d/rsh to get it "open" enough.

Now that's starting to sound like a wiki article! :D
pub_bronx wrote: Wiki article about net-installing an Octane from a FreeBSD box added here !

Nice! Thanks!

I remember seeing their CTO, Eng Lim Goh, demo something running on IRIX that very closely resembled today's Google Earth back in 2002. It blew everybody in the audience away, but it was only about two minutes of a 45 minute talk about "VANs" or "Visual Area Networks," which were being pushed as the tech that would change everything for SGI. During the cocktail hour, everyone was buzzing about the map demo while doubting that VANs would have any place in their organizations in the foreseeable future. Signal, meet noise.
cicero wrote: It's a quite old powermac 400 with PCI gfx that I use in the office for surfing, ssh shells, web radio, converting mac fonts, etc. No serious image/video/dtp applications. Maybe I should try to upgrade to tiger first. 10.3.9 is currently running on it.

Tiger runs great on that configuration. With RAM so cheap, make sure that you have at least 512 MB in the box. Personally, I suggest 1 GB, particularly on an older machine.
cicero wrote: Yes some days ago I upgraded to 768 MB. I found some SDRAM DIMMs in a dusty WinNT box ;)
Today I 've created some business cards in xpress. it runs quite well on this old box. But I'll keep my eyes open for additional memory. 1 GB is max?

768 MB is pretty good. As for how much RAM your machine can take, take a look at the appropriate machine page or the RAM configurator at .
kramlq wrote: So you should always have a large-ish swap file. Its difficult to say what size is best, but 1-1.5 or even 2 times the size of RAM is a common setting. Contrary to popular belief, this is purely as a heuristic - there is no way to calculate the exact amount of swap you need based on the amount of RAM you have installed.

Agreed. The amount of swap needed has almost nothing to do with how much RAM is installed; it has everything to do with how the system is used.

I've seen 4 GB systems work just fine with only 256 MB of swap, as long as they were used as general purpose workstations, and they weren't being used to manipulate multiple multi-GB images simultaneously. In my last shop, we generally set swap to 512 MB on IRIX 6.5 systems with drives 9 GB or smaller, and 1 GB on systems with drives 18 GB and larger. Never had to think about it, never ran into a problem. Older systems often had 256 or even 128 MB swap.

By the way, Mac OS X dynamically allocates swap space as swap files. I'm typing this on a MacBook Pro with 2 GB physical RAM under Tiger, while running six Firefox tabs, a full instance of XP Pro via Parallels, a couple of terminal windows, and Lotus Notes. Current swap = 137 MB.
Hi toxygen,
I used similar methods to what you described. While those are a lot of Opera tabs, my bet for the source of the difference between our systems is Photoshop, which has a lot of memory management tricks that it performs it is launched. It could also be that with 1 GB RAM, your system needs to start paging to swap earlier than a 2 GB system.
SAQ wrote: TextWrangler is a must in my opinion. Stock systems don't come with much in the way of ASCII editors.

...or make sure that you've installed X11 (it should be on your MacOS disks as an optional install package) and then download nedit for that good, old-fashioned, chewy, IRIX-flavored, text-editing goodness. Now available with special MacOS X Intel binary packaging!
..and, just for fun, a few screen savers might be cool.
Just came across a list of 16 Killer Apps to Put On Your Brand New Christmas Mac .

As a bonus, it has a picture of Alessandra Ambrosio (a Victoria's Secret model) with some Mac hardware.
kitsune wrote: a good place to pick up mac up hw and other stuff is otherworldcomputing. they have great prices and support is top-notch. load that mbp up with 4gb of ram for around $100. =)

I have to agree with that recommendation. I've been buying stuff from them for years, and they've always been great, including when I've had to return faulty hardware. They're not necessarily the cheapest, but they're usually close enough. (I don't know if they do any biz outside of the US & Canada.)
regan_russell wrote: What do you mean by Developer's Toolkit CD-ROM?

Perhaps he means the old "Developer's Tool box " discs, which included example code and documentation for a range of projects. Once upon a time, Developer's Program members could subscribe (for a fee) to receive DT cds a couple of times a year, and there also was a fair amount of stuff at the old DT website , which seems lost to antiquity.

{edited for clarity}
bjames wrote: I want to move my 6 drive SCSI Raid to my O2 from my Octane. Is there an quick easy way to do this? like copy a file over that contains the mount info or something to that effect? rather than configuring the drives from scratch? The drives have been optimized to handle uncompressed video.

If you're using XLV, it's pretty easy. What I'm about to say is from memory, since I don't have an xlv configured at the moment, so some of the detail might be off, but I've done this a couple of times without a problem.

Basically, there is only one xlv-specific change that you'll need to make. Everything else is standard IRIX, like creating a mount point and an entry in your Octane's /etc/fstab.

First, make sure that the XLV software is installed and running on the new system. It isn't part of the default IRIX installation. Use chkconfig to make sure that the xlv script runs when the system boots (as root or via su/sudo, "chkconfig xlv on").

When you power up an xlv-enabled system, it will probe the SCSI buses and try to identify attached xlv sets. It doesn't care if you move xlv member disks from one SCSI bus or another, as it assembles xlvs based on unique identifiers in the disk header, not by position on a SCSI bus. Once it identifies the xlv elements, it creates a device file, like /dev/xlv/mycrazyxlv. This is where it gets amazing. The disk headers of xlv disks contain hostname information. If the new system has the same hostname as the old system, and you've added the appropriate mounting information to /etc/fstab, everything should just work , no matter what SCSI bus or disk id the xlv disks now have.

If the xlv disks have a hostname in their headers which is different from the hostname of the system they are attached to, they can still be assembled, but they will now have a different device name with the following form: /dev/xlv/ Original_XLV_Hostname. mycrazyxlv.

Since your Octane probably has a different hostname from your O2, you'll need to make one of two possible adjustments. You can either change the hostname information on the xlv disks (preferred if you plan on keeping the xlv attached to the Octane) or update the device name in /etc/fstab (preferred if you plan on moving the xlv set back and forth between other systems).

To change the hostname information on the xlv disks, run xlv_mgr, and use the "change nodename" expression to update the xlv. For example, within xlv_mgr, you would use

Code: Select all

change nodename NEW_HOSTNAME mycrazyxlv
where NEW_HOSTNAME is your Octane's hostname, and mycrazyxlv is the xlv name. Note that you can execute this command on either the O2 or the Octane.

Consult the man pages for xlv_mgr and xlv_assemble for more information.

Oh, and, as always, if at all possible, make a backup of your filesystems before undertaking disk management operations. ;)