Linux 3.3 Released 314
diegocg writes "Linux 3.3 has been released. The changes include the merge of kernel code from the Android project. There is also support for a new architecture (TI C6X), much improved balancing and the ability to restripe between different RAID profiles in Btrfs, and several network improvements: a virtual switch implementation (Open vSwitch) designed for virtualization scenarios, a faster and more scalable alternative to the 'bonding' driver, a configurable limit to the transmission queue of the network devices to fight bufferbloat, a network priority control group and per-cgroup TCP buffer limits. There are also many small features and new drivers and fixes. Here's the full changelog."
Keep it up. (Score:5, Insightful)
Re:Keep it up. (Score:5, Insightful)
Great! (Score:1, Insightful)
Now how many of these features are out of beta and actually work?
Re:Way to go....... (Score:5, Insightful)
Re:Great timing (Score:4, Insightful)
Probably never will be either. Its usefulness was always questionable at best anyway, it is a GREAT academic exercise that I'm interested in just cause I've been developing my own 'x64 os' as a learning experience so the tactics they use I like to learn about, from a practical perspective as a system admin, its silly.
Mission critical infrastructure where you would want continuous availability is running on a cluster which can stand to have a host rebooted for upgrades so live splicing kernels is pointless in those situations.
ksplice is for people in moms basement who want an uptime long penis, not for anyone who actually needs service availability.
ksplice is a treatment for a symptom, which has a long list of side effects that are non-obvious to your non-developer sysadmins, which means most.
Clusters are the vaccination/condom that prevents you from developing the problem in the first place
Re:Way to go....... (Score:4, Insightful)
It's the people who purposefully comment on threads about products they DON'T like that create a problem.
I wouldn't say it's a problem. They, like to the people who like it, are simply stating their opinions.
Re:Great timing (Score:4, Insightful)
And it's doable, you didn't give good requirements, but in your CCTV example, all you need is to store the data on a SAN.
Basically if your design depends on the fact that none of your servers will go down, then you need a new designer.
Re:Great timing (Score:5, Insightful)
Really?
Well, explain this to me:
When you have 8 servers each with 2 PCI-E Quad E1 Digium Cards, handling a total of 248 inbound calls on toll free numbers, with an average of 200 simultaneous channels per server 24/7, how do you cluster that? When you have analog CCTV cameras running into 4 servers each with 16 channels of video, well, how do you cluster that?
Not everything is HTTP over TCP/IP. Not everything is easily solved with a load-balancing reverse proxy or DNS balancing/failover. Not everything can be clustered. In those situations, and I speak from experience (those two are real-life situations I deal with), not rebooting is real fucking important. And you have two options: either you leave systems unpatched and wait for the next 5 minutes downtime window that might be a year from now, hoping nothing bad happens, or you live patch those motherfuckers.
If that wasn't enough, I can mention at least an extra 20 cases where clustering isn't an option, and neither is rebooting.
The explanation is fairly simple. Your hardware does not meet the redundancy and failover requirements for the uptime expectations you've set. Equipment designed for extensive uptimes and critical services have built-in redundancy. Take your phone "server" for example. We run several types of phone service, and the one most similar to what you describe runs as a pair of servers, each with a primary and redundant connection. If either server fails, or if the connected switches/routers fail, they can failover to the backup hardware without even interrupting a call in progress. When we need to upgrade the servers, we do them one at a time.
Rebooting always has to be an option. Always. Why? Because sooner or later you're going to have hardware failure, and you'll be rebooting whether or not you want to. Going with your 'head in the sand' approach only means your customers will feel a much greater impact from the inevitable downtime than they would if you'd properly designed your systems in the first place.
Re:Way to go....... (Score:5, Insightful)
So... which part of this release actually provides a compelling reason to use Linux over any other OS?
You've been itching for something to run on that TI C6X system you built?
The fanboisim here makes me gag. Apple has nothing on you guys.
Hey Cowboy did you know that the Linux Kernel currently runs most smart tvs, bd players, and other home entertainment devices. I would be willing to bet that the number of Samsung, Sony, LG and other tvs and devices running on the Linux kernel is much greater than the number of Macs, and PC currently in use combined!
The reason for this is that any manufacturer can use OpenSource software like the Linux kernel and modify it to their own needs without sending money to Redmond for every device they sell. This is why Microsoft and Apple have failed in the embedded market with perhaps the exception of some car companies like Ford Motors. Ballmer can rant, rave and do all the paid shill crap he wants. Fact is as the kernel becomes more open to modification from companies like Google with Android optimisations and slick coding Microsoft will become irrelevant in many markets.
The post was about the most important core software released in history so go pound on your PC, and post how linux sucks somewhere where someone cares. The Linux kernel is one hell of allot more that just the base of an OS as you perceive it.
Re:Way to go....... (Score:1, Insightful)
>It's the people who purposefully comment on threads about products they DON'T like that create a problem.
So what are your thoughts on how Slashdot has used a broken window to represent Windows stories for at leats a decade?
And what are your thoughts about how until just a few months ago Slashdot used the borg-ified face of the world's largest philanthropist as an icon for Microsoft stories?
Are you (and all the people who gave your post karma) implying that Slashdot itself and all its editors are douchebags who create problems?
Because I would agree with that sentiment.
Re:Way to go....... (Score:2, Insightful)
Re:I suppose I have to start building... (Score:4, Insightful)
You gotta wonder what the hell Linus is thinking on this.
Well, while he's a hard nail on code quality he's always been a pragmatic man. When it's an interface used on hundreds of millions of Android devices it's something worth supporting if he can do it as long as it doesn't interact badly with the mainline code. And that's exactly why something like wakelocks are still out while others are in. I don't think Linus believes in the one perfect system, if he has to support different IPCs then fine but maybe the implementation can share code and work towards supporting several approaches.
Remember it's not in anybody's interest to diverge just to diverge, it's just that sometimes it's better to do your own thing and show that it works rather than trying to get permission to change an old recipe. A lot of branches have lived in parallel to mainline and eventually gotten merged in as the real needs and differences - not just the NIH and semantics - have emerged. Getting over these hurdles and keeping the kernel from fracturing into smaller branches that each go their separate ways has always one of the true strengths of the project.
Re:Android? As in Google? As in NSA spyware? (Score:4, Insightful)
Re:Way to go....... (Score:5, Insightful)
Re:Way to go....... (Score:4, Insightful)
Re:Way to go....... (Score:3, Insightful)
And what constitutes an "asshole" is subjective. I don't see anything wrong with stating your opinion and doing nothing besides that, even if someone thinks it's harsh.
Re:Way to go....... (Score:2, Insightful)
Don't forget the 250+ million Android devices floating around out there.
Linux is everywhere these days although mostly invisible (some would call that "seamless").