Language:
switch to room list switch to menu My folders
Go to page: First ... 72 73 74 75 [76] 77 78 79 80 ... Last
[#] Thu Feb 10 2022 17:07:26 EST from Nurb432

[Reply] [ReplyQuoted] [Headers] [Print]

Never used it, but i knew it existed. 



[#] Sat Feb 12 2022 10:48:19 EST from IGnatius T Foobar

[Reply] [ReplyQuoted] [Headers] [Print]

It's pretty cool for what I'm doing. As you know, I have my servers sitting behind a VPN that places them onto the public Internet in a location other than the physical location of the hardware. But I want to be able to bypass that and go directly in, so my activity doesn't take that path.

So now I can go from my desktop --> host system --> router vm --> destination vm , without having to stop and type another SSH command at each hop. After my previous message I played around with the "ProxyJump" directive in .ssh/config which made it even easier; now I can just type a command to SSH to the server I want and the client figures out the jump hosts to use to get there. Very cool!

[#] Mon Feb 14 2022 19:17:29 EST from LoanShark

Subject: Re: Where are they now?!

[Reply] [ReplyQuoted] [Headers] [Print]

I never got into BSDI but I heard it had some devoted fans.

Well, that was a commercial BSD. I remember my local ISP used it back in the dialup days in Maine.

[#] Tue Feb 15 2022 18:55:48 EST from IGnatius T Foobar

Subject: Re: Where are they now?!

[Reply] [ReplyQuoted] [Headers] [Print]

The big one over here (Cloud 9 Internet) relied heavily on BSDI. They came into existence before Linux was really a thing, so it made sense for them and a lot of other early ISPs at the time.

Now that dialup is dead, most of the dialups that survived live on as IT service providers, some as small web hosting companies. Cloud 9 shuttered its data center and moved its hosting infrastructure into a cage inside a local colocation provider (no extra points for guessing which one). WestNet seems to have been reduced to nothing more than an email service, presumably for hangers-on who didn't want to change their addresses. There was a third one around here, but I can't remember its name so I don't know what became of them, if anything. [A

[#] Wed Mar 30 2022 04:46:56 EDT from xfalkenx

Subject: Re: Where are they now?!

[Reply] [ReplyQuoted] [Headers] [Print]

 

What I can say is that I suffered, as an IT crew, the last days of HP/UX and IBM AIX and their respectives HP9000 and RS/6000 architectures.

 

Application development was pushing hard with an unstoppable thirsty of memory, io and CPU and these architectures had already grew to their maximum capabilites. Then daily and nightly crashes came. JVM ran out of memory. Oracle ran out of memory. The operating systems ran out of semaphores. Locks everywhere. Crashes and more crashes. This situation lasted up to 2010-2012.

 

Then, Linux and Intel came by multiplying per 10 the memory, the CPU. Also the virtualisation was largely implemented in the form of Xen or VMware. No more crashes. Then peaceful nights came and I eventually could watch a full movie at the cinema with no oracle.restart.sh interruptions.

 

I experienced that twice in my work life in two very different sectors.

 


[#] Thu Mar 31 2022 23:12:20 EDT from IGnatius T Foobar

Subject: Re: Where are they now?!

[Reply] [ReplyQuoted] [Headers] [Print]

Heh. AIX is still around, running on POWER10, but I'm sure most of those machines are running "System i" to keep the legacy AS/400 apps running. This is funny: the current AIX web site says "Build cloud native apps and co-locate with AIX." As if anyone is going to go through the trouble of containerizing their applications and then run them on an overgrown RS/6000 instead of commodity hardware.

There was a time when even underpowered Unix Iron could outperform newer x86 hardware, but those days are long gone. Let us raise a glass to the memory of the good ones, and to the demise of the awful ones.

[#] Fri Apr 01 2022 07:13:11 EDT from Nurb432

Subject: Re: Where are they now?!

[Reply] [ReplyQuoted] [Headers] [Print]

Agreed, we still have one in the back room at the main data center. And a 2nd out on our DR site  ( we used to have small ones spread out into many of the remote offices.. those days are gone now )

Tho to be fair, its down to 2 apps now and they are working to get them off and onto 'generic' platforms by the end of the year....

Thu Mar 31 2022 11:12:20 PM EDT from IGnatius T Foobar Subject: Re: Where are they now?!
Heh. AIX is still around, running on POWER10, but I'm sure most of those machines are running "System i" to keep the legacy AS/400 apps running. This is funny: the current AIX web site says "Build cloud native apps and co-locate with AIX." As if anyone is going to go through the trouble of containerizing their applications and then run them on an overgrown RS/6000 instead of commodity hardware.

There was a time when even underpowered Unix Iron could outperform newer x86 hardware, but those days are long gone. Let us raise a glass to the memory of the good ones, and to the demise of the awful ones.

 



[#] Sat Apr 02 2022 13:28:43 EDT from xfalkenx

Subject: Re: Where are they now?!

[Reply] [ReplyQuoted] [Headers] [Print]

I totally agree, IG. On the other hand, ex workmates who I am still in touch with defend the POWER and AIX tandem with a stallmanish fervour.

I managed POWER virtualisation on z-Series for AIX and Suse Linux guests. It was magic on 2004 to see popping up a new CPU in top real time but that HMC tool was tedious to use (we used it through an X terminal IIRC) and nowadays the most basic Xen or Proxmox environment performs equal or better with more modern technologies (Ceph, clustering...) and, important, for less euros.

The same on Fujitsu's Primepower and Sun's LD.

Those Old Guard machines fell on the nostalgia side. It's not bad. We have feelings!

 



[#] Tue Apr 05 2022 09:07:48 EDT from IGnatius T Foobar

Subject: Re: Where are they now?!

[Reply] [ReplyQuoted] [Headers] [Print]

The POWER architecture *is* pretty awesome. It's just too expensive for mainstream computing at this point.

[#] Tue Apr 05 2022 13:23:35 EDT from Nurb432

Subject: Re: Where are they now?!

[Reply] [ReplyQuoted] [Headers] [Print]

Which is a shame. I was trilled when Apple put it in desktops.  ( ya i do realize PowerPC is a subset of Power, but still in the same family.. )

Tue Apr 05 2022 09:07:48 AM EDT from IGnatius T Foobar Subject: Re: Where are they now?!
The POWER architecture *is* pretty awesome. It's just too expensive for mainstream computing at this point.

 



[#] Fri Apr 08 2022 10:51:16 EDT from xfalkenx

Subject: Re: Where are they now?!

[Reply] [ReplyQuoted] [Headers] [Print]

And Amiga with PPC boards which made us call the first Pentium processors "Lentium" which, in Catalan, lent means slow.

 

Tue Apr 05 2022 13:23:35 EDTfrom Nurb432 Subject: Re: Where are they now?!

Which is a shame. I was trilled when Apple put it in desktops.  ( ya i do realize PowerPC is a subset of Power, but still in the same family.. )



[#] Mon Apr 18 2022 09:20:34 EDT from IGnatius T Foobar

Subject: Re: Where are they now?!

[Reply] [ReplyQuoted] [Headers] [Print]

i do realize PowerPC is a subset of Power, but still in the same

A subset perhaps, but Linux treated POWER and PowerPC as the same architecture.

I'm going to take an educated guess that the difference consisted of instructions and registers that only mattered to AS/400 workloads. And that's fine, of course; no one wanted to run AS/400 software on a Power Mac (although that would be really funny).

[#] Sat Aug 13 2022 19:18:01 EDT from zelgomer

[Reply] [ReplyQuoted] [Headers] [Print]

smartmontools. Could they have made this any more impenetrable? I have to spend 5 minutes reading the man pages every time I just want to look at the state of my disks, and then I have to spend another 20 minutes googling what the hell these parameters mean (is a bigger value good or bad?). And I think I just discovered that this whole time I've had it setup to, instead of mailing me warnings, invoke a script which calls run-parts on an empty dir, effectively doing nothing.

[#] Sat Aug 13 2022 19:21:40 EDT from zelgomer

[Reply] [ReplyQuoted] [Headers] [Print]

2022-08-13 23:18 from zelgomer <zelgomer@uncensored.citadel.org>
smartmontools. Could they have made this any more impenetrable? I have

to spend 5 minutes reading the man pages every time I just want to look

at the state of my disks, and then I have to spend another 20 minutes

googling what the hell these parameters mean (is a bigger value good or

bad?). And I think I just discovered that this whole time I've had it

setup to, instead of mailing me warnings, invoke a script which calls

run-parts on an empty dir, effectively doing nothing.



Oh yeah, and similar subject: Network UPS Tools (NUT). Exact same story. I'm not completely confident I have it configured correctly, and when I want to look at anything I first spend 5 minutes searching installed packages and the binaries they deploy because fuck if I can ever remember these command names. upsc? upscmd? upsmon??!

[#] Sat Aug 13 2022 19:23:54 EDT from zelgomer

[Reply] [ReplyQuoted] [Headers] [Print]

2022-08-13 23:21 from zelgomer <zelgomer@uncensored.citadel.org>
2022-08-13 23:18 from zelgomer <zelgomer@uncensored.citadel.org>

smartmontools. Could they have made this any more impenetrable? I
have

to spend 5 minutes reading the man pages every time I just want to
look

at the state of my disks, and then I have to spend another 20 minutes


googling what the hell these parameters mean (is a bigger value good

or

bad?). And I think I just discovered that this whole time I've had it


setup to, instead of mailing me warnings, invoke a script which calls


run-parts on an empty dir, effectively doing nothing.



Oh yeah, and similar subject: Network UPS Tools (NUT). Exact same
story. I'm not completely confident I have it configured correctly, and

when I
want to look at anything I first spend 5 minutes searching
installed packages and the binaries they deploy because fuck if I can

ever remember these command names. upsc? upscmd? upsmon??!



I guess if I weren't a complete moron I would take the time to make a cheat sheet for myself. Sorry, last double post.

[#] Sat Aug 13 2022 19:51:44 EDT from Nurb432

[Reply] [ReplyQuoted] [Headers] [Print]

I do that every time i do something once in a blue moon.

then i can never find it.. and have to fight with it again...:(

Sat Aug 13 2022 07:23:54 PM EDT from zelgomer
I guess if I weren't a complete moron I would take the time to make a cheat sheet for myself. Sorry, last double post.

 



[#] Mon Aug 22 2022 11:49:56 EDT from IGnatius T Foobar

[Reply] [ReplyQuoted] [Headers] [Print]


Linux network namespaces (or as the rest of the world calls them ... VRFs)

It's annoying that when you move an interface from the default namespace into a different one, it loses its configuration. It is maddening that the "ip link set <interface> netns <namespace>" command doesn't have a "preserve addresses" option.

wtf, linux? This seems like something people would want to do on a regular basis.

[#] Mon Aug 22 2022 13:38:49 EDT from Nurb432

[Reply] [ReplyQuoted] [Headers] [Print]

Its what happens when you have a guy who was still in school that started the thing, and a chaotic bazaar development model.  i will admit that it has got better over the years, unless you want to talk about people like pottering infecting things :P 

Consolation with things like FreeBSD is its far more organized, and stupid crap rarely gets thru committee.   But yes, it does slow things down too, so its a trade-off.

IMon Aug 22 2022 11:49:56 AM EDT from IGnatius T Foobar


wtf, linux? This seems like something people would want to do on a regular basis.

 



[#] Mon Aug 22 2022 17:32:17 EDT from LoanShark

[Reply] [ReplyQuoted] [Headers] [Print]

wtf, linux? This seems like something people would want to do on a
regular basis.

I've never needed/wanted to do anything like this.

Basically it sounds like you're tinkering with something low-level that's normally handled by containerd and/or docker.

[#] Wed Aug 24 2022 09:19:42 EDT from IGnatius T Foobar

[Reply] [ReplyQuoted] [Headers] [Print]

Normally I'd agree with that sentiment. In this case ... I have a virtual machine acting as a VPN router for this site and the rest of my Internet-facing servers. Right now it's using that classic trick of setting the default route to go through the tunnel, and then a /32 static route making the VPN connection itself go to the "real" gateway.

The WireGuard web site describes [ https://www.wireguard.com/netns/ ] some neat tricks you can do by moving its interface into a different namespace, while the tunnel provider remains in the namespace in which it was created.
This can be used for containers, by configuring a tunnel and then making the wg0 (or whatever) interface the container's only network. They also show a way to move the tunnel provider into its own namespace and then make wg0 the default namespace's only network, avoiding the trick of having to set up a /32 to the tunnel provider.

I want to do something similar, even though my tunnel happens to be L2TP instead of WireGuard.

The long and short of it is that I want the machine that runs the router to continue to accept incoming connections on its Ethernet interface, but this cannot happen when its default gateway is set to an existing tunnel.

I've been experimenting with the idea of running L2TP in the default namespace and then moving ppp0 (or whatever) into its own namespace, along with the inside network. This doesn't work because ppp0 loses its configuration when it is moved into the other namespace. I might try it again. I was doing the move inside an ip-up script, and that script *does* know the addresses of both ends of the connection when it starts, so it might work if I set them again after the move.

Go to page: First ... 72 73 74 75 [76] 77 78 79 80 ... Last