geekhack
geekhack Community => Other Geeky Stuff => Topic started by: _rubik on Tue, 30 October 2018, 10:45:20
-
I'm curious to hear GH's thoughts on this Red Hat acquisition.
Personally, I'm a little bummed out. I don't see that IBM has anything to add to Red Hat, so there's nowhere but down. Best care scenario, I see RH being left alone for a few years before IBM makes a move.
With RH's acquisition of CoreOS and IBM's acquisition of Red Hat, IBM now has a choke hold on virtual server space and container orchestration.
-
Don't like it. Red Hat is probably going down in the long term. IBM has not been a good company in recent years.
BTW. There is some previous discussion on the topic in What Linux Distro do the Linux users of GH use? (https://geekhack.org/index.php?topic=56515.msg2674132#msg2674132).
-
Don't like it. Red Hat is probably going down in the long term. IBM has not been a good company in recent years.
BTW. There is some previous discussion on the topic in What Linux Distro do the Linux users of GH use? (https://geekhack.org/index.php?topic=56515.msg2674132#msg2674132).
Indeed there was. Figured it was a little off topic and deserved its own thread. People seem pretty upset about it -- more so than when MS bought Github
-
IBM = NSA
-
You think Red Hat software is going to get backdoors in the future?
-
Honestly... with IBMs ethics, I don't see why not.
I also don't want to lean into conspiracy theories either, and there's no way the REHL community would let that fly.
-
Some interesting observations from Jon "maddog" Hall (http://www.linux-magazine.com/Online/Blogs/Paw-Prints-Writings-of-the-maddog/IBM-Purchase-of-Red-Hat-Software-There-is-no-fear-except-fear-itself-with-thanks-to-FDR).
-
Some interesting observations from Jon "maddog" Hall (http://www.linux-magazine.com/Online/Blogs/Paw-Prints-Writings-of-the-maddog/IBM-Purchase-of-Red-Hat-Software-There-is-no-fear-except-fear-itself-with-thanks-to-FDR).
My favorite read so far https://purpleidea.com/blog/2018/10/30/the-end-of-a-red-giant/
-
I also don't want to lean into conspiracy theories either, and there's no way the REHL community would let that fly.
"Conspiracy theories" have got a bad name. With how the US "works" many things are not really as far-fetched as they may seem. Some things are even done in the open without people reacting.
-
Some interesting observations from Jon "maddog" Hall (http://www.linux-magazine.com/Online/Blogs/Paw-Prints-Writings-of-the-maddog/IBM-Purchase-of-Red-Hat-Software-There-is-no-fear-except-fear-itself-with-thanks-to-FDR).
My favorite read so far https://purpleidea.com/blog/2018/10/30/the-end-of-a-red-giant/
"... and they can instead focus on building good products and saving IBM" - says a lot.
Wonder if it will be renamed Blue Hat Linux :))
-
It seems the running joke is that at least Oracle did not acquire Redhat.
-
It's about RedHat + OpenShift so IBM can make a PaaS play to their enterprise customers.
-
What effects, if any, will the IBM acquisition of RedHat have on Fedora and/or CentOS?
-
I doubt it will be an issue.
There's an old saying in the industry: No-one gets fired for going with IBM.
-
What effects, if any, will the IBM acquisition of RedHat have on Fedora and/or CentOS?
Most of the servers at work plus my new(er) home server run CentOS, so I hope little, if any, impact.
-
What effects, if any, will the IBM acquisition of RedHat have on Fedora and/or CentOS?
Fedora is the least of their worries, so I think it'll largely remain the same. As for RHEL and it's CentOS counterpart, I wish I could say the same.
I don't exactly see them changing the OS' themselves around too much (maybe RHEL's payments) but the greater issue is the monopoly they're slowly building on the enterprise Linux spaces (distros, containers, and virtualization).
-
IBM naturally points out the positives with the acquisition.
https://developer.ibm.com/blogs/2018/10/28/ibm-and-red-hat-are-coming-together-and-developers-are-the-winners/
-
The company that killed Linux with systemd is killed by the last big Unix company. Ha!
-
The company that killed Linux with systemd is killed by the last big Unix company. Ha!
Last big Unix company that hasn't been acquired by another ... Oracle still sells Solaris, latest release came out this year and I think it has at least more mind-share than AIX, if not a larger install-base still.
-
The latest HP-UX version is from 2017, so we could assume it is still "active" as well (and nobody will ever buy HP, I guess). But yes, it is the last big Unix company that hasn't been acquired by another. ;)
-
The company that killed Linux with systemd is killed by the last big Unix company. Ha!
I'd love your honest opinion on why systemd killed Linux. Indexing binary logs and boot-time service startup is fast as all hell.
I guess you could make the inflexible monolith argument, but I see it as a good thing for enterprise distros.
-
Binary logs are broken logs. You can't just grep them.
-
Binary logs are broken logs. You can't just grep them.
Oh boy. This is the most stupid argument against them. If you have more than a handful of machines, you're not going to grep logs. If you do, you're doing something terribly wrong. And even if you only care about a single machine, grepping logs is still a terribly inefficient way to work with them. Binary logs can provide way more protection against tampering, corruption and whatnot than plain text logs do. Not to mention that most of the time, you'll compress logs older than a day or two anyway, and boom, you have binary data, which you have to uncompress first to grep them. Not much different than using journalctl.
(Not saying the Journal is good - it isn't -, just that binary logs are not the reason it's bad.)
Sorry for the off-topic, but this silly argument gets me triggered all the time. I spent too much time working in the logging industry, I guess.
I'm curious to hear GH's thoughts on this Red Hat acquisition.
I worked for a company once that got acquired by IBM. We were promised to be able to remain independent, that IBM won't interfere with our day to day lives. That was a lie. Pretty much everything we built had to be rebuilt on top of IBM technologies, even if they were completely rubbish. We had to switch to Lotus Notes (oh dear god, that thing is a disaster!) for email and calendaring. Conference tickets, hotel bookings and other things had to be cancelled (and in some cases, paid for by employees when they could not be cancelled), even though they were approved previously by management (and this happened because another acquired company in the same unit happened to underperform badly! nothing to do with us).
IBM also had a very strong propaganda and push for software patents, and some crazy policies like having to ask management for every single open source contribution one might do, even on their free time. Higher ups could give blanket approvals, but that placed additional risk on them, and wasn't a practice IBM was recommending. Oh, thet list had to be printed. If you were a frequent contributor to open source software, you were easily looking at hundreds of pages. They eventually backed down when a few of us sent them a few hundred pages of PDFs with our lists, but it took quite an effort and a lot of push to get even this far.
Needless to say, I left IBM pretty soon after. It's not a good place to work. They may tout they like Linux, and they do: they want to reap all the benefits of free software, but contribute a little back as possible. Embrace, extend, extinguish. That's the plan down the road, and always has been.
In short, the acquisition will be great for RedHat shareholders. It will be great for the brand (both RedHat and IBM). I'm fairly sure it will turn out to be a disaster for free software and open source folks within RedHat.
-
Binary logs can provide way more protection against tampering, corruption and whatnot than plain text logs do.
Huh? Signed log files?
I did a search on systemd and found a bunch of reports on systemd logging getting corrupted and just stopping.
-
Binary logs are broken logs. You can't just grep them.
Oh boy. This is the most stupid argument against them. If you have more than a handful of machines, you're not going to grep logs. If you do, you're doing something terribly wrong. And even if you only care about a single machine, grepping logs is still a terribly inefficient way to work with them.
This is the key point here. Security and corruption issues aside, binary logs provide an efficient way to query service logs.
Conference tickets, hotel bookings and other things had to be cancelled (and in some cases, paid for by employees when they could not be cancelled), even though they were approved previously by management (and this happened because another acquired company in the same unit happened to underperform badly! nothing to do with us).
For whatever reason, this is a problem at every large company I've been at. You would think that they would support learning relevant trade skills and sharing them with co-workers, but they would rather have you in the salt mines. The cynical side of me says that they would sooner higher new-grads with new found knowledge and pay them half the salary than to educate already experienced folks.
IBM also had a very strong propaganda and push for software patents, and some crazy policies like having to ask management for every single open source contribution one might do, even on their free time. Higher ups could give blanket approvals, but that placed additional risk on them, and wasn't a practice IBM was recommending. Oh, thet list had to be printed
:vomit: Better to ask for forgiveness than permission? Ether way, a toxic environment when engineers are penalized for innovation outside of company time/walls.
-
Binary logs can provide way more protection against tampering, corruption and whatnot than plain text logs do.
Huh? Signed log files?
I did a search on systemd and found a bunch of reports on systemd logging getting corrupted and just stopping.
Not talking about the journal, but binary logs in general. The journal is bad. But it isn't bad because binary logs. Nevertheless, the journal does support signed logs, and when journal files get corrupted, most of the time it will just rotate to use a new one. It used to crash hard way back when, but those days are mostly over I think. The rotating away behaviour is already better than text logs, where discovering corruption isn't so trivial, and most programs will just keep on writing.
Text logs get corrupt quite often too, mind you, they're less frequently discovered because it's hard to catch the corruption by the human eye, grep will helpfully not care if it finds garbage, and so on. So many ways to paper over the problem.
When implemented properly, both binary and text logs are append-only. They can get corrupted under pretty much the same conditions.
-
:vomit: Better to ask for forgiveness than permission? Ether way, a toxic environment when engineers are penalized for innovation outside of company time/walls.
They are rewarded for innovation, as long as that results in a patent. No patent, no love.
(Yep, that's even worse.)
-
I'll just echo chamber some of the podcast things I've heard recently: the general population can now point to the Red Hat Acquisition and say that's proof that working with and developing open source software actually does pay (~$34B).
And yet the part of me that sometimes peaks at pictures of the Red Hat headquarters--that part of me is somewhat doleful that Red Hat is now part of a much larger corporate entity.
Anyway, Fedora Workstation would be my first choice for hidpi if I hit a wall with Arch. I hope the acquisition won't kill the desktop innovation.