Language:
switch to room list switch to menu My folders
Go to page: First ... 58 59 60 61 [62]
[#] Wed Aug 26 2020 13:53:13 EDT from IGnatius T Foobar

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


Fun discovery of the day ... at least for me.

cp(1), or at least the version of cp found on Linux/Linux, has a relatively new option "--reflink" to create a copy-on-write (COW) copy when used on a filesystem that supports it (such as btrfs).

This is going to be really useful for creating low-overhead clones of things without having to create subvolumes and snapshots. Right away I'm going to use this to build groups of virtual machines, because I'm sure btrfs can do "linked clones" better than VMware can.

The btrfs people and the NetApp people should get together and standardize an NFS command to do this remotely.

[#] Wed Aug 26 2020 17:22:10 EDT from LoanShark

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


I wonder if that plays well with overlayfs

[#] Wed Aug 26 2020 17:52:36 EDT from IGnatius T Foobar

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

I don't know. From what I can tell, when the option first appeared, it was just "--reflink" and the man page said "works with btrfs". Recently it changed [ https://linux.die.net/man/1/cp ] to "--reflink=<when>" and "<when>" can be "always" or "auto" with the default being "always" if neither option is specified, and btrfs is no longer mentioned, so they're obviously trying to generalize it.

It looks like they're trying to make the COW-copy heuristics as close as they can to sparse file heuristics, and probably generalizing the kernel interface for it too.

[#] Sun Dec 13 2020 13:29:46 EST from zooer

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

[#] Mon Dec 14 2020 13:13:52 EST from IGnatius T Foobar

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

Typical IBM. It's actually amazing that they haven't fired everyone in Raleigh and moved the job of ruining Red Hat to a bunch of people in India yet. In the mean time, the original lead developer of CentOS has already begun work on a new distribution: "Rocky Linux" with the same goal of providing a free (gratis) Linux operating system that is binary compatible with the Red Hat distribution of the same version number.

This is a big stake in the heart of Red Hat Linux. IBM clearly does not understand that enterprise Linux users *want* to use the exact same operating system in both free-unsupported and paid-supported versions. Ubuntu got it right: give away the bits for free, sell support to those who need it.

[#] Mon Dec 14 2020 19:15:31 EST from Nurb432

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

Nah, they know, they just dont care. 

Pay up or GTFO !



[#] Tue Dec 15 2020 00:01:26 EST from ParanoidDelusions

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

At one point, after doing the analysis of cost, and consulting with a CIO from a sister company I decided to go with an IBM XIV instead of sticking with EMC sans, which weren't scaling well, and were costing us a fortune. The XIV was far cheaper - and is actually an incredible platform for shared storage arrays. 

But when the service contract came back up - I understood why it was so inexpensive for so much storage. 


Mon Dec 14 2020 19:15:31 EST from Nurb432 @ Uncensored

Nah, they know, they just dont care. 

Pay up or GTFO !



 



[#] Tue Dec 15 2020 00:20:25 EST from IGnatius T Foobar

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

We switched from EMC Symmetrix to IBM 2105 ESS (the "Shark") for similar reasons.
Just like with their mainframe stuff, they make sure you upgrade when they feel the platform has completed its run by sharply raising the maintenance costs. Old school mainframe hands know this by the name "the migration path."

Unfortunately for them, when workloads went virtual, changing storage vendors is just a live-migration away; you don't even need to stop the running workload to do it. We went from IBM to a mix of NetApp and EMC Clariion without any customers even knowing or caring (once we got out of the bad habit of writing vendor names and equipment models into the contracts). Of course, nowadays, "software-defined storage" is all the rage, which is just a fancy way of saying "oh, maybe we should actually use the disk drive bays in all these servers, maybe spread the data around so we don't need a giant channel and yoyobytes of cache."

[#] Tue Dec 15 2020 00:23:07 EST from ParanoidDelusions

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

I am definitely out of fashion on the latest trends in designer Enterprise Architecture fashion. :) Haven't cared since about 2014. 

Tue Dec 15 2020 00:20:25 EST from IGnatius T Foobar @ Uncensored
 "software-defined storage" is all the rage, which is just a fancy way of saying "oh, maybe we should actually use the disk drive bays in all these servers, maybe spread the data around so we don't need a giant channel and yoyobytes of cache."

 



[#] Fri Dec 18 2020 12:30:57 EST from Ragnar Danneskjold

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

My future son in law works on the Z-Series...... It's amazing how backward compatible it is, but also able to embrace the future.

IBM got it right there, but somehow can't do the same elsewhere.

[#] Fri Dec 18 2020 14:03:48 EST from IGnatius T Foobar

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

Well, who knows ... maybe a de-emphasis on x86 will encourage more adoption of mainframes (running Linux, not Z/OS) in high end environments that just need massive compute. A cloud in a box. You can probably run millions of containers on a Z.

"This mainframe here runs our massive enterprise application suite! It handles millions of transactions per second."

"Oh, is it written in COBOL?"

"Naah, we just whacked most of it together in Node.js"

[#] Sat Dec 19 2020 15:20:51 EST from Nurb432

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

I have found that pricing keeps the big boxes out of a lot of possible markets. Unless you *need* the transactional rates the big iron can give you, you dont want to pay the cost.  Hell its one reason we are all sitting here on a 'personal computing device' now..  IBM wanted to squeeze that cash, and ended up creating the market that marginalized themselves.



[#] Thu Dec 24 2020 09:19:36 EST from Nurb432

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

*shudder*

Thu Dec 24 2020 06:54:09 EST from vegangd Subject: How To Install NMAP On Ubuntu 20.04 LTS and Other Distro

 snap



 



[#] Thu Dec 24 2020 12:20:08 EST from IGnatius T Foobar

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

I'm considering that posting to be spam and have deleted it.



[#] Sat Jan 16 2021 08:06:36 EST from Nurb432

Subject: Debian to drop 32bit x86

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

Well, that is a shame. I understand why, but its still sad to see it go and i think that its premature considering its continued heavy use. 

 

https://www.theregister.com/AMP/2021/01/14/debian_bullseye/



[#] Sat Jan 16 2021 16:19:26 EST from IGnatius T Foobar

Subject: Re: Debian to drop 32bit x86

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

It had to happen at some point. For that matter, eventually 32-bit x86 has to get dropped by the kernel too. Backwards compatibility is great but it comes at the cost of complexity and the effort required to maintain it. It would be interesting to see a Linux system boot with the kernel and initrd both stored in the EFI system partition, on a legacy-free system with no BIOS or 32-bit support.

The system you are logged into right now is running on a 32-bit Debian VM.
I haven't gone to 64-bit because of the effort required to convert my database.
I really ought to get that done.

[#] Sun Jan 17 2021 07:42:23 EST from Nurb432

Subject: Re: Debian to drop 32bit x86

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

Ya i also understand why, but as a 'end user' it still sucks.

I have a couple of old laptops that will be dead at that point. ( i dont believe video drivers exist on netbsd ).  And while i have not had them out of the closet for a long time, and dont 'need' them, i just hate when good hardware becomes unusable for not fault of its own. Have a couple of ATOM boxes in there too, unsure if they are 32 or 64. Used to use them as low power servers until i started buying those lenovo mini I5's

 

Sat Jan 16 2021 16:19:26 EST from IGnatius T Foobar Subject: Re: Debian to drop 32bit x86
It had to happen at some point. For that matter, eventually 32-bit x86 has to get dropped by the kernel too. Backwards compatibility is great but it comes at the cost of complexity and the effort required to maintain it. It would be interesting to see a Linux system boot with the kernel and initrd both stored in the EFI system partition, on a legacy-free system with no BIOS or 32-bit support.

The system you are logged into right now is running on a 32-bit Debian VM.
I haven't gone to 64-bit because of the effort required to convert my database.
I really ought to get that done.

 



[#] Sun Jan 17 2021 12:25:39 EST from ParanoidDelusions

Subject: Re: Debian to drop 32bit x86

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

As someone who runs an Amiga with SSL that can connect via ethernet to IP based sites... 

Those old retro machines won't be dead at that point. They'll become supported through unofficial methods. 

In fact as they become truly retro because of lack of mainstream support - as they become more rare because people dispose and recycle their "unsupportable" hardware - they're likely to become more valuable with people who prize being able to run 32bit code native on original hardware rather than FPGA or emulation. 

Especially iconic examples like Thinkpads that capture a "moment in time" experience. 

 

Sun Jan 17 2021 07:42:23 EST from Nurb432 Subject: Re: Debian to drop 32bit x86

Ya i also understand why, but as a 'end user' it still sucks.

I have a couple of old laptops that will be dead at that point. ( i dont believe video drivers exist on netbsd ).  And while i have not had them out of the closet for a long time, and dont 'need' them, i just hate when good hardware becomes unusable for not fault of its own. Have a couple of ATOM boxes in there too, unsure if they are 32 or 64. Used to use them as low power servers until i started buying those lenovo mini I5's

 

Sat Jan 16 2021 16:19:26 EST from IGnatius T Foobar Subject: Re: Debian to drop 32bit x86
It had to happen at some point. For that matter, eventually 32-bit x86 has to get dropped by the kernel too. Backwards compatibility is great but it comes at the cost of complexity and the effort required to maintain it. It would be interesting to see a Linux system boot with the kernel and initrd both stored in the EFI system partition, on a legacy-free system with no BIOS or 32-bit support.

The system you are logged into right now is running on a 32-bit Debian VM.
I haven't gone to 64-bit because of the effort required to convert my database.
I really ought to get that done.

 



 



[#] Mon Jan 18 2021 06:24:14 EST from darknetuser

Subject: Re: Debian to drop 32bit x86

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

2021-01-16 08:06 from Nurb432
Subject: Debian to drop 32bit x86
Well, that is a shame. I understand why, but its still sad to see it

go and i think that its premature considering its continued heavy
use. 

 

https://www.theregister.com/AMP/2021/01/14/debian_bullseye/


Everybody is trying to get 32 bits offed.

Oracle'sJDK no longer supports 32 bit x86, nuff said.

I agree it sucks because there are lots of things that still require 32 bit, and a lot of people does not want to run a multiarchitecture os.

[#] Mon Jan 18 2021 06:28:54 EST from darknetuser

Subject: Re: Debian to drop 32bit x86

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

2021-01-17 07:42 from Nurb432
Subject: Re: Debian to drop 32bit x86
Ya i also understand why, but as a 'end user' it still sucks.

I have a couple of old laptops that will be dead at that point. ( i

dont believe video drivers exist on netbsd ).  And while i have not

had them out of the closet for a long time, and dont 'need' them, i

just hate when good hardware becomes unusable for not fault of its

own. Have a couple of ATOM boxes in there too, unsure if they are 32

or 64. Used to use them as low power servers until i started buying

those lenovo mini I5's

 

Still you can get other distributions for x86 32 bit for now.

When Tails dropped 32 bit, it screwed me hard because at the time I only had access to old computers, as do some people I know.

I think dropping things that are in use in the wild because "nobody uses them" is such a first worlder thing to do. I remember when people was claiming to stop serving Linux distributions in CDs because it was time to migrate to DVD or USB images, disregarding the fact the difference in cost still matters for a whole lot of people.

Not everybody is a greedy Californian bastard with access to piles of DVDs and optic fibre.

Go to page: First ... 58 59 60 61 [62]