You get it on Darwin as well, was paying around with a Rhapsody build and it came up, also tried it on verbose OSX after messing with the time on the NVRAM
I could imagine a programmer in the 70s typing that, thinking "Okay, if people are still messing around with this trash in the 2020s, then more than anything else, that's just sad. Also it might break in the code so we have to say something." This was a time when the moon landings were only a couple years back, and that we'd have 2001ASO-style tech in 2001.
Turns out, it wasn't. Imagine the security implications of having a cron job that downloads a plaintext hostfile over telnet. That would have been a considerable security risk when the Internet consisted of a few thousand nodes, it would be completely untenable for the modern Internet. (hell, I didn't even like using hostfiles for my home network of 4 nodes, so now I have a DNS server, less a security reason than laziness since it's populated by the DHCP server, actually, I think they might be the same thing in that implementation)
I remember a time when there was a web page at the University of Illinois Urbana-Champaign which the National Center for Supercomputing Applications (NCSA) kept up to date with "all" the web sites, or at least all new .edu and .gov sites as universities and government labs began putting up new hand-coded web pages. This was in the days of the NCSA Mosaic browser, when search options were pretty minimal and even before companies like Microsoft or IBM had hung up their shingles on the world wide web. Obviously maintaining the page eventually turned into a fool's errand, but I don't know when they stopped updating it.
@@StarkRG i don't see it as more risky than normal dns. If you signed the file you could test that its correct too unlike regular in era dns where you wouldn't know if the packets were messed with or even what server the dns request really got routed to. Its certainly not less secure than ftp downloading executables in the era. Remember we used to do banking over phone lines and also with just unsecured plain telnet.
@@lassikinnunen Signed it how? Not with PGP which didn't exist until 1991. RSA existed but wasn't really used until the late 80s. Messing with packets isn't nearly as easy as gaining access to a server and editing a text file. I'm not saying DNS is considerably more secure than downloading a hostfile, but it _is_ slightly more secure.
Remember when Solaris 8 came out and while it was installing, you could browse the web? It asked all the questions it needed uo front and didnt bother you until it was done. Many hours where used to browse the onion.
Love seeing you dig deep into the guts of all the stuff I mucked with in ye olden days. Some times I miss the old days. Then I remember NIS existed, and then I don’t feel so bad.
My first major task as a professional admin was to convert our infrastructure from SunOS 4.1.4 to Solaris. This video brought back a lot of memories. Weirdly, even though we used to use NIS for passwords and groups, I never knew there was a NIS-only version of Netscape. At least we never used NIS for host resolution and always used DNS. We also briefly flirted with NIS+ but wound up going with LDAP.
We had a weird Sysadmin guy back in the early 00s who had NIS on a (production) Gentoo linux box and a Samba sever on an old SGI machine. He lost his shit because he couldn't sync the NIS on the linux box with the NIS on the SGI.
Long ago I looked into migrating from NIS to NIS+ on some AIX servers (with mixed clients, from Solaris to AIX to Linux and even HP-UX). I even briefly set up an NIS+ domain on some new machines. The phrase "bag of hurt" hadn't been coined yet or I would certainly have used it. I stayed with NIS.
And Perl was a well fledged language by this time. All they had to do to make this work is get a few guys to figure out all the weird ways people write config files and spit out ones that work with nis+!! Yea. That didn't happen in the sun world either. Remember webmin? That could take a config file from a service 13 people actually used and webify it.
Had to work on a couple NIS network system a few years back, and it was neat watching you go through the same stuff I had to. We also used a lot of LDAP on systems that could be upgraded, which was on an enterprise system with replication. 9 times out of 10 it meant if you messed up you might have wiped everything across multiple networks... Good times.
I worked for Sun in the five years prior to the Oracle acquisition and as soon as I saw that first error message I burst out laughing (before spending the rest of the video crying of course). Of course there was some greybeard in sfbay who would grudgingly consent to his system having a graphical web browser put on it but he'd be damned if it meant using that upstart nameserver that didn't even need the local BOFH to add new sites to the hosts file.
What's worse is, I had to manage this horrendous stack with clients on AIX, HP-UX and Dynix/ptx as well. At some point our ypserver was on SunOS, then Solaris, and finally the ypserv ended up on AIX as we removed Sun hardware, but we also added Linux clients at that time too. Absolute horror show of an admin task and I'm glad to see the back of it.
4:46 - interesting. Have you had exposure to SGI's "Indigo Magic Desktop", 4Dwm, etc? I _much_ preferred those back in the day. And indeed, frankly miss them, even now.
Brings back bad memories as a sysadmin in the 1990’s. Long after everyone else used DNS by default to resolve host names, Sun used NIS by default, and it was not as reliable as DNS. And it took a bit of work to make Solaris act like a more normal UNIX and not use stupid NIS for name resolution.
Netscape 4.7 did not became the basis for firefox. Netscape-4.x codebase was actually old and hacky, so Netscape decided to start from scratch - this rewrite was actually what became Firefox in the long run.
In my experiance, no corporations really used nis+. By that time ldap was coming to be and corporations started to mive to that. Even though that has its own serious caveats even to this day. Most companies now that Ive seen use rsync for passwd,shadow and there sudo file. Its just easier to manage.
There was a huge rewrite of Netscape's interface for the unreleased Netscape 5, which is often unfairly blamed for the demise of the company even though it was the precise thing that enabled Firefox to be born. And even that rewrite didn't completely remove the old code, though I don't imagine a single line has remained unchanged.
Most of NSS and NSPR which handles a lot of the low level plumbing and portability came straight across. Parts of Gecko did as well, and this version recongizes `about:mozilla` and has a quote from the Book of Mozilla as well
Well, this took me back. While I was not directly involved in the NIS+ setup just after Y2K, I remember VERY well when it went down. It was eventually dropped like a hot potato ASAP, to be replaced by Sun's LDAP implementation. A step closer to sanity at least.
LDAP is a great idea on paper. Making it work cross platform is a nightmare. I mentioned in another comment that most places I've worked we just rsync passwd, sudo and shadow. With a little bit of scripting you can get a home directory set up. In some cases, I just dump them into a /tmp directory if they wouldn't need to save anything. If they need persistent data, NFS mount it.
Ho boy, I remember not only maintaining but also installing NIS (twice) at previous job. It worked usually fine....until it didn't. And debugging with tcpdump and Wireshark is a new skillset I am probably not gonna use anytime soon :D
Wireshark is still very relevant today. I use it for sip traffic analysis all the time. I use it for NFS and smb auth issues also. A friend of mine uses it for SDR also. That blew my mind
On one Sun box, I mistyped the NIS domain name during my initial Solaris install. I corrected that error by rerunning the installer (thankfully, that prompt was not very far in), but somehow it had gotten deep into some sort of undocumented nonvolatile memory, and that workstation would randomly start spewing error messages about not being able to locate the misspelled domain. Everything worked, it just sometimes spewed error messages. Directly to the console, not via syslog, at a rate of about one per second. It was otherwise harmless (everything still worked fine), but the workstation's user did not appreciate the console messages messing up his screen. I used all sorts of diagnostics to try and search all files on disk and examine the NVRAM for the misspelled domain, all to no avail. Never could find the mistyped string, though obviously it was there someplace. I discovered that the ypbind executable binary contained the string "/dev/console", realized that "/dev/null" has fewer characters than "/dev/console", so patched it so it babbled its error messages to /dev/null instead. That workstation ran with my hand-patched ypbind for years.
Thankfully NIS was mostly gone by the time I entered the industry. However that doesn't mean I'm free of NIS trauma. Solaris default automount would cause anything to do with /home to just be broken out of the box. Always had to disable automount on there before trying to create users or you'd be in for a world of pain
It's 2024 and I'm still trying to yank NIS out of a production network. Also, "Daemon" is pronounced the same as "Demon", it's an archaic spelling of the same word.
I'd be really interested in who you work for! This is kind of the jobs I look for, updating random weird boxes thrown together that all work and nobody remembers HOW. I'd like a job interview!
@@rnts08 if you're worried about all the background checks and stuff I've been through that. I could name the banks and firms. The exchanges. But let's not do that here.
Hey NCommander have you tried running windows on btrfs (yes not a typo) Its super jank and one of the steps needed is literally doing a live conversion from ntfs to btrfs... Do far only vm's have been done, maybe its time to take the step?
I believe SunOS has a separate file for name resolution. It's not in etc, either. You need to tell it to use etc/hosts specifically, it doesn't default to that. I remember struggling with this 20 years ago
Lovly I was wondering if u know of any good dns server for Windows NT 3.51 because by default I was only able to use dhcp I was able to make gopher, www, ftp work by installing IIS, but never found DNS Server snywhere
DNS Server for NT 3.51 was part of BackOffice - a purchasable set of server products to install on top of it. Still that too was a preview thing. So you are better off using a 3rd party DNS server like BIND or pfSense, or spin up an NT 4 box I think. :)
It was never as hard as you make out. NIS and NIS+ could sometimes cause issues, but always down to user error - good change management practice is always a necessary step for enterprise systems
Dont wanna to nitpick, but Netscape never was base of Firefox. Mozilla Suite was. And speaking of relation between Netscape and Mozilla suite its not that clear. Some claims Mozilla suite based on in-developement version of Netscape which was open-sourced, other say it was indeed Netscape code, but good people at Mozilla very quickly thrown away most of Netscape code, except certain more complex bits (which then haunted them for years).
@@NCommander It have about:mozilla because "Mozilla" actually was internal name for Netscape, way way prior to Mozilla Foundation. This is reason why every http useragents still have word Mozilla in it. Speaking of engine its also not so easy. Current Firefox engine isnt Gecko, its mostly Servo. What i try to convey is that you cant call human to be "son of fish" just because both is vertebrate.
I remember these rumors. I can't remember who it was but a Dev pointed out in code that large portions of the code where moved over from the engine. Can we put this one to bed guys.
@@djksfhakhaks Some code indeed was moved over, no objection on that. For example Mozilla struggled for years with "Mork" database - its internal Netscape format they invented and mozilla devs called it "completely mindf*cked", absolutely impossible to understand and very prone to failure if any changed applied. This was reason they adopted Sqlite - to get rid of that. At time they worked on that it was real popcorn show to read their bugzilla. Well, so long, Netscape code.
I refuse to stand for unga-bunga network diagnostics being slandered in such a way. If the switch doesn't respond to _clonk_, it's PFE didn't deserve to live.
I like the boot message "preposterous time in file system" - guess nobody at SUN expected installs 30 years after release
You get it on Darwin as well, was paying around with a Rhapsody build and it came up, also tried it on verbose OSX after messing with the time on the NVRAM
That message predates Sun. I think it goes all the way back to unix v7.
I could imagine a programmer in the 70s typing that, thinking "Okay, if people are still messing around with this trash in the 2020s, then more than anything else, that's just sad. Also it might break in the code so we have to say something."
This was a time when the moon landings were only a couple years back, and that we'd have 2001ASO-style tech in 2001.
@@kargaroc386 maybe there where people entering fake dates as a joke ?
I thought this was an April Fool's joke. I can't imagine HTML-era Internet without DNS ever being practical.
Turns out, it wasn't. Imagine the security implications of having a cron job that downloads a plaintext hostfile over telnet. That would have been a considerable security risk when the Internet consisted of a few thousand nodes, it would be completely untenable for the modern Internet. (hell, I didn't even like using hostfiles for my home network of 4 nodes, so now I have a DNS server, less a security reason than laziness since it's populated by the DHCP server, actually, I think they might be the same thing in that implementation)
I remember a time when there was a web page at the University of Illinois Urbana-Champaign which the National Center for Supercomputing Applications (NCSA) kept up to date with "all" the web sites, or at least all new .edu and .gov sites as universities and government labs began putting up new hand-coded web pages. This was in the days of the NCSA Mosaic browser, when search options were pretty minimal and even before companies like Microsoft or IBM had hung up their shingles on the world wide web. Obviously maintaining the page eventually turned into a fool's errand, but I don't know when they stopped updating it.
@@StarkRG i don't see it as more risky than normal dns.
If you signed the file you could test that its correct too unlike regular in era dns where you wouldn't know if the packets were messed with or even what server the dns request really got routed to. Its certainly not less secure than ftp downloading executables in the era.
Remember we used to do banking over phone lines and also with just unsecured plain telnet.
@@lassikinnunen Signed it how? Not with PGP which didn't exist until 1991. RSA existed but wasn't really used until the late 80s. Messing with packets isn't nearly as easy as gaining access to a server and editing a text file. I'm not saying DNS is considerably more secure than downloading a hostfile, but it _is_ slightly more secure.
@@lassikinnunen Essentially, MITM has always been an overrated threat. It does exist, certainly, but sniffing is generally a larger risk.
Remember when Solaris 8 came out and while it was installing, you could browse the web? It asked all the questions it needed uo front and didnt bother you until it was done. Many hours where used to browse the onion.
Love seeing you dig deep into the guts of all the stuff I mucked with in ye olden days. Some times I miss the old days. Then I remember NIS existed, and then I don’t feel so bad.
My first major task as a professional admin was to convert our infrastructure from SunOS 4.1.4 to Solaris. This video brought back a lot of memories. Weirdly, even though we used to use NIS for passwords and groups, I never knew there was a NIS-only version of Netscape. At least we never used NIS for host resolution and always used DNS. We also briefly flirted with NIS+ but wound up going with LDAP.
We had a weird Sysadmin guy back in the early 00s who had NIS on a (production) Gentoo linux box and a Samba sever on an old SGI machine. He lost his shit because he couldn't sync the NIS on the linux box with the NIS on the SGI.
"Horrific, Painful and DNS"
Great to see the Horrific part, next Painful? (NIS+)
Beat me to it. Nisplus….shudder
Long ago I looked into migrating from NIS to NIS+ on some AIX servers (with mixed clients, from Solaris to AIX to Linux and even HP-UX). I even briefly set up an NIS+ domain on some new machines. The phrase "bag of hurt" hadn't been coined yet or I would certainly have used it. I stayed with NIS.
And Perl was a well fledged language by this time. All they had to do to make this work is get a few guys to figure out all the weird ways people write config files and spit out ones that work with nis+!! Yea. That didn't happen in the sun world either.
Remember webmin? That could take a config file from a service 13 people actually used and webify it.
Had to work on a couple NIS network system a few years back, and it was neat watching you go through the same stuff I had to. We also used a lot of LDAP on systems that could be upgraded, which was on an enterprise system with replication. 9 times out of 10 it meant if you messed up you might have wiped everything across multiple networks... Good times.
I worked for Sun in the five years prior to the Oracle acquisition and as soon as I saw that first error message I burst out laughing (before spending the rest of the video crying of course). Of course there was some greybeard in sfbay who would grudgingly consent to his system having a graphical web browser put on it but he'd be damned if it meant using that upstart nameserver that didn't even need the local BOFH to add new sites to the hosts file.
I am happy to see you're back. I've been playing your Doom video on repeat for the last year of waiting! :D
What a trip down memory lane this is. Thank you.
What's worse is, I had to manage this horrendous stack with clients on AIX, HP-UX and Dynix/ptx as well. At some point our ypserver was on SunOS, then Solaris, and finally the ypserv ended up on AIX as we removed Sun hardware, but we also added Linux clients at that time too. Absolute horror show of an admin task and I'm glad to see the back of it.
Man, the duct tape must have been epic.
The PTSD kicked in as soon as you said NIS
Oh man, thanks so much for the nostalgia trip. This was basically the first decade or so of my IT career right there.
Your background noise made me think I had backlight bleed for a second
When I first started Linux as a hobby I think yp was one of those services the Mandrake manual warned me wasn't really secure.
4:46 - interesting. Have you had exposure to SGI's "Indigo Magic Desktop", 4Dwm, etc? I _much_ preferred those back in the day. And indeed, frankly miss them, even now.
Brings back bad memories as a sysadmin in the 1990’s. Long after everyone else used DNS by default to resolve host names, Sun used NIS by default, and it was not as reliable as DNS. And it took a bit of work to make Solaris act like a more normal UNIX and not use stupid NIS for name resolution.
This was nice to see
Ahhhhh welcome back! It’s been a while!
Netscape 4.7 did not became the basis for firefox. Netscape-4.x codebase was actually old and hacky, so Netscape decided to start from scratch - this rewrite was actually what became Firefox in the long run.
In my experiance, no corporations really used nis+. By that time ldap was coming to be and corporations started to mive to that. Even though that has its own serious caveats even to this day. Most companies now that Ive seen use rsync for passwd,shadow and there sudo file. Its just easier to manage.
I wonder if there's any Netscape code still left in Firefox? That would be really neat.
There was a huge rewrite of Netscape's interface for the unreleased Netscape 5, which is often unfairly blamed for the demise of the company even though it was the precise thing that enabled Firefox to be born. And even that rewrite didn't completely remove the old code, though I don't imagine a single line has remained unchanged.
Most of NSS and NSPR which handles a lot of the low level plumbing and portability came straight across. Parts of Gecko did as well, and this version recongizes `about:mozilla` and has a quote from the Book of Mozilla as well
I'm getting more into syadmin stuff, and watching this just made me glad that NIS/NIS+ is barely/if at all used... Jesus Christ, you poor soul.
Well, this took me back. While I was not directly involved in the NIS+ setup just after Y2K, I remember VERY well when it went down.
It was eventually dropped like a hot potato ASAP, to be replaced by Sun's LDAP implementation. A step closer to sanity at least.
LDAP is a great idea on paper. Making it work cross platform is a nightmare. I mentioned in another comment that most places I've worked we just rsync passwd, sudo and shadow. With a little bit of scripting you can get a home directory set up. In some cases, I just dump them into a /tmp directory if they wouldn't need to save anything. If they need persistent data, NFS mount it.
Hooray new video!
Ho boy, I remember not only maintaining but also installing NIS (twice) at previous job. It worked usually fine....until it didn't. And debugging with tcpdump and Wireshark is a new skillset I am probably not gonna use anytime soon :D
Wireshark is still very relevant today. I use it for sip traffic analysis all the time. I use it for NFS and smb auth issues also. A friend of mine uses it for SDR also. That blew my mind
On one Sun box, I mistyped the NIS domain name during my initial Solaris install. I corrected that error by rerunning the installer (thankfully, that prompt was not very far in), but somehow it had gotten deep into some sort of undocumented nonvolatile memory, and that workstation would randomly start spewing error messages about not being able to locate the misspelled domain. Everything worked, it just sometimes spewed error messages. Directly to the console, not via syslog, at a rate of about one per second. It was otherwise harmless (everything still worked fine), but the workstation's user did not appreciate the console messages messing up his screen. I used all sorts of diagnostics to try and search all files on disk and examine the NVRAM for the misspelled domain, all to no avail. Never could find the mistyped string, though obviously it was there someplace. I discovered that the ypbind executable binary contained the string "/dev/console", realized that "/dev/null" has fewer characters than "/dev/console", so patched it so it babbled its error messages to /dev/null instead. That workstation ran with my hand-patched ypbind for years.
What do you use for Sun workstation emulation?
qemu afaik
Look at the title bar, it says QEMU
We're so back
Thankfully NIS was mostly gone by the time I entered the industry. However that doesn't mean I'm free of NIS trauma. Solaris default automount would cause anything to do with /home to just be broken out of the box. Always had to disable automount on there before trying to create users or you'd be in for a world of pain
It's 2024 and I'm still trying to yank NIS out of a production network. Also, "Daemon" is pronounced the same as "Demon", it's an archaic spelling of the same word.
I'd be really interested in who you work for! This is kind of the jobs I look for, updating random weird boxes thrown together that all work and nobody remembers HOW. I'd like a job interview!
@@djksfhakhaks I, uh, am actually hiring pretty soon...
@@djksfhakhaks try banks and insurance companies.
@@rnts08 if you're worried about all the background checks and stuff I've been through that. I could name the banks and firms. The exchanges. But let's not do that here.
if you need a freelancer, I actually do that kind of work
Hey NCommander have you tried running windows on btrfs (yes not a typo)
Its super jank and one of the steps needed is literally doing a live conversion from ntfs to btrfs...
Do far only vm's have been done, maybe its time to take the step?
I believe SunOS has a separate file for name resolution. It's not in etc, either. You need to tell it to use etc/hosts specifically, it doesn't default to that. I remember struggling with this 20 years ago
Just out of curiosity, where do you get binaries for Netscape on SunOS 4?
Cursed? Wait until you deal with Sendmail.
mmmm M4 macros
Lovly
I was wondering if u know of any good dns server for Windows NT 3.51 because by default I was only able to use dhcp
I was able to make gopher, www, ftp work by installing IIS, but never found DNS Server snywhere
DNS Server for NT 3.51 was part of BackOffice - a purchasable set of server products to install on top of it. Still that too was a preview thing. So you are better off using a 3rd party DNS server like BIND or pfSense, or spin up an NT 4 box I think. :)
You need the NT 3.51 Resource Kit to get MS DNS.
Hes back!
Ooooo, ooh! Do UUCP next?!
one thing you can say about NIS is that it was way easier to get single sign-on between computers working than LDAP.
oh also i still love OpenLook a lot, i wish somebody would make a gtk theme based on it for modern gtks.
True that.
It was never as hard as you make out. NIS and NIS+ could sometimes cause issues, but always down to user error - good change management practice is always a necessary step for enterprise systems
Good change management can be copying flat files. The computer is made by humans, it can be made better
Dont wanna to nitpick, but Netscape never was base of Firefox. Mozilla Suite was. And speaking of relation between Netscape and Mozilla suite its not that clear. Some claims Mozilla suite based on in-developement version of Netscape which was open-sourced, other say it was indeed Netscape code, but good people at Mozilla very quickly thrown away most of Netscape code, except certain more complex bits (which then haunted them for years).
This version of Netscape has about:mozilla in it, and the underlying engine is still Gecko. Close enough for me.
@@NCommander It have about:mozilla because "Mozilla" actually was internal name for Netscape, way way prior to Mozilla Foundation. This is reason why every http useragents still have word Mozilla in it. Speaking of engine its also not so easy. Current Firefox engine isnt Gecko, its mostly Servo.
What i try to convey is that you cant call human to be "son of fish" just because both is vertebrate.
I remember these rumors. I can't remember who it was but a Dev pointed out in code that large portions of the code where moved over from the engine. Can we put this one to bed guys.
@@djksfhakhaks Some code indeed was moved over, no objection on that. For example Mozilla struggled for years with "Mork" database - its internal Netscape format they invented and mozilla devs called it "completely mindf*cked", absolutely impossible to understand and very prone to failure if any changed applied. This was reason they adopted Sqlite - to get rid of that. At time they worked on that it was real popcorn show to read their bugzilla.
Well, so long, Netscape code.
@@Vednier I have a feeling he knows more about it than you.
7:11: Ah, the first generation of what would be today called zero clients...
I refuse to stand for unga-bunga network diagnostics being slandered in such a way. If the switch doesn't respond to _clonk_, it's PFE didn't deserve to live.
Thanks for at nice video.
the colors are nice though
"helped" tie networks together, he says…
5:15: That's Igg-DRAZZ-le if you're Norse-ty...
YOGdrasil 5:14
I really miss netscape browser. I wish there was a new version of this mozzlia, I think it sucks.
Yes. I miss typing a URL, hitting enter and having a smoke break while it loaded. Those where the good ol days!😅
Isn't SeaMonkey a continuation of the original Mozilla Project?
waiting ...
...that's what the status bar will show 😜