FreeBSD was once "the power to server" but in an AWS world we have fallen way, waaay behind and there seems no interest to fix it!

[PMc]

I have experienced first hand a senior manager who, upon learning of a catastrophic bug (multixact in v9.4 (I think)) in postgres was absolutely catatonic when he found out we didn't have paid support. He actually said "Well, who can I blame then? You? Or you?..." (as his eyes moved around the room ... it was hilarious).

Yeah. Any idea how to fix that?
(From what I understand it is not to be fixed in software, neither in contracts or business economics. It is a problem of the mind-state of the business-people, and should be fixed in the programming of their mind. But how?)

 

[Mjölnir]

Like security, independence is an illusion; it's an utopian, idealistic concept. It simply doesn't exist in real life.

 

[kpedersen]

He actually said "Well, who can I blame then? You? Or you?...

Yes, very good point. I have come across my fair share of this. I think sometimes it also rubs off on some engineers (potentially when they want to enter management and get promotion). They would choose a tool that they can't fix themselves and then just sit there blaming the vendor why they cannot achieve something. Unfortunately, it does very little to solve the problem! The deadline will still not be met. The requirement will still not be achieved.

I especially dislike this kind of management when we cannot be an early adopter of a technology because we have to wait for this slow monolith of a tool to catch up rather than simply implementing the support ourselves. Tools such as Unity3D and Microsoft's C compiler are the worst offenders here.

I even see Docker as a bit like this. The kids won't try anything new or interesting until there is a Docker container on that stupid consumer DockerHub which appears 6 months too late. Just so they can blame the packager if something goes wrong? Daft.

 

[Deleted member 30996]

Yeah. Any idea how to fix that?
(From what I understand it is not to be fixed in software, neither in contracts or business economics. It is a problem of the mind-state of the business-people, and should be fixed in the programming of their mind. But how?)

How gratifying for a Programmer to see and hear people talking about the programming and de-programming of wetware. I'm back in vogue after banishment of that training in my field since 1982.

...they said they chose Linux because that's what they used in school.

Remember the controversy over their use in schools when pocket calculators were invented? I quit right after that but know where the story went from there.

I'm always the odd one out. FreeBSD never gives me anything to complain about.

 

[PMc]

How gratifying for a Programmer to see and hear people talking about the programming and de-programming of wetware.

It's to the point: vonNeumann's architecture and Luhmann's systems-theory together allow us to create analytic descriptions of things which formerly were considered "magick". How many of those concerned did realize they where re-defining Gnosis?

I'm back in vogue after banishment of that training in my field since 1982.

Yeah, that was about the time when it was no longer cool living as a hobo.

 

[Deleted member 30996]

Yeah, that was about the time when it was no longer cool living as a hobo.

There was a homeless guy who posted here when I first joined but I believe he asked for his account be deactivated.

 

[8bitGlitch]

Yes, very good point. I have come across my fair share of this. I think sometimes it also rubs off on some engineers (potentially when they want to enter management and get promotion). They would choose a tool that they can't fix themselves and then just sit there blaming the vendor why they cannot achieve something. Unfortunately, it does very little to solve the problem! The deadline will still not be met. The requirement will still not be achieved.

I especially dislike this kind of management when we cannot be an early adopter of a technology because we have to wait for this slow monolith of a tool to catch up rather than simply implementing the support ourselves. Tools such as Unity3D and Microsoft's C compiler are the worst offenders here.

I even see Docker as a bit like this. The kids won't try anything new or interesting until there is a Docker container on that stupid consumer DockerHub which appears 6 months too late. Just so they can blame the packager if something goes wrong? Daft.

I hear this argument all the time. I wanted to introduce, Ansible, Cacti and Nagios into the mix at a gig for a cost effective solution to manage a network sometime back. However, I was told "who do we call for support, yell at and blame? Oh... I will just blame you". I was like Mr. Tim the Tool Man Taylor "AeUhhhh???" In addition, I hear it from other folks as well, when they try to promote Debian vs Ubuntu.

For my personal items, I have packed up and left CentOSville for good for FreeBSD, so my journey has just begun. Started out with C128, moved to Amiga around 1993 to play Hart of China (still have my original copy and my Amiga 'they were cheap when they went out of business') and one other game I can't find. Heck... I was still typing college papers on my Amiga way into 2002 until I got forced not to do so, since University only accepted MS Word file format. Do not think I had a windows machine until 2004-5. At work always used HP-UX or in one case Solaris, but since I am network, really never cared what OS my workstation was, since I lived on hardware or embedded OS devices.

 

[gschadow]

Hi, I realize that this thread had been rudely hijacked. I think this forum is pretty useless for real FreeBSD support, and whoever cares can take this as feedback also.

Meanwhile on the FreeBSD-performance email list the replies were more useful.

The freebsd-performance February 2021 Archive by thread

Of course there is no sudden magic answer but less denial and useless off topic chatter.

 

[kpedersen]

From the mailing list:

There are some commercial FreeBSD products running on aws . Maybe the vendors know some stuff that can help ?

Sounds pretty similar to the advice from ralphbsz to be fair.
But now you know the correct place where to go to report things such as this in future

 

[8bitGlitch]

Hi, I realize that this thread had been rudely hijacked. I think this forum is pretty useless for real FreeBSD support, and whoever cares can take this as feedback also.

Meanwhile on the FreeBSD-performance email list the replies were more useful.

The freebsd-performance February 2021 Archive by thread

Of course there is no sudden magic answer but less denial and useless off topic chatter.

AWS... who? Why even worry about running FreeBSD at high performance in AWS just scale out with multiple installs... or roll your own service. I would think FreeBSD in AWS playground free-for-all would be a niche install base or user base. And one that would have insight into how things work behind the scenes at AWS to fine tune FreeBSD image to get anywhere close to Amazon Linux speeds. I say, just use Amazon Linux when you have to play in AWS, since you know it most likely is the BEST fined tuned for their service. Plus managmenet would love you... for giving the blame all on Amazon... ++smiles++

I have not had the opportunity to build anything in production for AWS for clients. It becomes too expensive with data costs, and the nickle and dime to death. I have done test pilots with hardware and firewall appliences for demarks and vpn in side of vpn, but it gets messy fast.

 

[scottro]

Yeah, the denial part is bad, but you can get a lot of help here.
Unfortunately, whenever there's an instance of Linux doing better in some circumstance there is this tendency to get into FreeBSD is better, it's someone else's fault. (And hey, we all know Amazon is evil). As for off-topic, it's the ole bikeshed thing, but I think almost any forum is prone to that. Mailing lists are often a bit stricter and therefore, sometimes more beneficial. Plus the mailing lists usually have more developers on them.

Frequently Asked Questions for FreeBSD 12.X and 13.X

Frequently Asked Questions, and answers, covering all aspects of FreeBSD

docs.freebsd.org

But these forums are also incredibly useful, with most people, including the moderators, ready and willing to help.

 

[richardtoohey2]

these forums are also incredibly useful, with most people, including the moderators, ready and willing to help.

Definitely!

People keep coming here and asking questions for which the bulk of people on here (mostly users and power users?) don't have answers.

There seems to be a perception that these forums are the place to get any and every question answered and issues resolved. But that doesn't seem to be the case, leading to frustration and provocative titles including things like "there seems to be no interest to fix it" - which will rile some people.

I think everyone on these forums has an interest in things being fixed, improved, kept-up-to-date. I think we'd all (well mostly!) agree that would be great if every printer, every USB device, every application, every piece of conceivable hardware, and every cloud providers' offering worked with and performed well on FreeBSD and worked as well as Linux, Windows, etc. Wouldn't it be nice if all providers thought FreeBSD should have the same level of support as Linux & Windows?

But (a) FreeBSd doesn't have the same market-share (b) FreeBSD doesn't have the same resources and (c) these forums don't seem to be the place to make things happen. You'll get help and support here but I don't think it's a direct line to key developers or movers & shakers in the FreeBSD world (I don't know where that is, though!)

Back on topic - this was good to see - there's definitely interest and desire to make things better:

Meanwhile on the FreeBSD-performance email list the replies were more useful.

The freebsd-performance February 2021 Archive by thread
Of course there is no sudden magic answer

 

[mark_j]

Definitely!

People keep coming here and asking questions for which the bulk of people on here (mostly users and power users?) don't have answers.

There seems to be a perception that these forums are the place to get any and every question answered and issues resolved. But that doesn't seem to be the case, leading to frustration and provocative titles including things like "there seems to be no interest to fix it" - which will rile some people.

I think everyone on these forums has an interest in things being fixed, improved, kept-up-to-date. I think we'd all (well mostly!) agree that would be great if every printer, every USB device, every application, every piece of conceivable hardware, and every cloud providers' offering worked with and performed well on FreeBSD and worked as well as Linux, Windows, etc. Wouldn't it be nice if all providers thought FreeBSD should have the same level of support as Linux & Windows?

But (a) FreeBSd doesn't have the same market-share (b) FreeBSD doesn't have the same resources and (c) these forums don't seem to be the place to make things happen. You'll get help and support here but I don't think it's a direct line to key developers or movers & shakers in the FreeBSD world (I don't know where that is, though!)

Back on topic - this was good to see - there's definitely interest and desire to make things better:

Hear Hear!


Many people also forget the age old saying "You get what you pay for". Never has this been truer of open source software.

FreeBSD is free, Amazon is not. Why the OP never approached/pressured AWS is beyond me. They're the ones offering the service and charging the money and yet FreeBSD forums are copping the grief? AWS want to offer FreeBSD? Let them pay to fix the issues.

Whether FreeBSD is at fault or not, AWS is the one charging money, so chase them. If they offer a substandard service, stop paying them and move on. If it means they stop offering FreeBSD, so be it. FreeBSD isn't getting anything out of it, is it?

YOU GET WHAT YOU PAY FOR.

 

[reddy]

People will come and offer different perspectives and also introduce relevant small talk from time to time, this is the difference between a forum and a mailing list. As far as I am concerned, I strongly feel that if the OP wants to blame someone for the amount of small talk this thread contains compared to the average on this forum, he should start by looking at the title of this thread. It is too easy to talk down to people, and dismiss the many helpful answers that have been offered without looking in the mirror.

 

[drhowarddrfine]

In the meantime...

 

[Beastie7]

That's a whole lot of what I paid for.

 

[Snurg]

I think it is plausible to assume that Amazon, being aware that FreeBSD 13 will be focused on performance improvements, might have decided to postpone the next AWS FreeBSD optimization overhaul until the 13 release.
I mean, how much sense would it make to invest much in optimizations that would be obsoleted very soon?

And to add a bit OT, I cannot help but scratch my head and find it a bit strange, how coincidently, just immediately before Amazon banned Parler, a costly AWS data leak opened up.

 

[Deleted member 66267]

Maybe I would consider donating a small cash to the FreeBSD Foundation so now I'm contributed, I will now have the ability to blame them.

 

[richardtoohey2]

Maybe I would consider donating a small cash to the FreeBSD Foundation so now I'm contributed, I will now have the ability to blame them.

A small cash donation entitles you to request one documentation fix.

A few milion dollars will help with bigger issues.

 

[mark_j]

People will come and offer different perspectives and also introduce relevant small talk from time to time, this is the difference between a forum and a mailing list. As far as I am concerned, I strongly feel that if the OP wants to blame someone for the amount of small talk this thread contains compared to the average on this forum, he should start by looking at the title of this thread. It is too easy to talk down to people, and dismiss the many helpful answers that have been offered without looking in the mirror.

You're absolutely correct.
Forums are not contracted support areas. You must expect some off-topic, especially when the answer/solution has been posted and the OP continues to ignore this.
You're also totally correct in that the OP's headline title was inflammatory at best.

 

[mark_j]

In the meantime...

I never went through the benchmarks, but isn't beta built with WITNESS? If so, like all things benchmarking, believe only your own experience.
I'm also surprised such a systemd/linux fan-boy site like that has FreeBSD benchmarks, unless they're there to prove the l33tn355 of leenoox.

 

[tuaris]

Or, take the typical 80A fuse for a 20-flat house, with every flat having an individual fusing of 32A.
This only works well due to the simultaneity factor. If only a small number of flats would draw their full rated current, it would result in a brownout or even blackout for all.

Anyway, so "cloud" in reality means, it allows me to sell you 4 real amperes for the price of 32A

That's the best analogy I've ever heard to describe 'the cloud'. Mind if I use it?

I'm (obviously) no fan of AWS and the like, but Amazon Linux and many other Linux distributions are optimized for the AWS platform. It has drivers and functions that are written to work better with the underlying virtual hardware. FreeBSD on the other hand currently doesn't, and that's why you see lower performance.

A great example is the gettimeofday system call which is known to be much slower on AWS's EC2. FreeBSD doesn't have a user space gettimeofday system call for AWS in the same way it was implemented for Azure. Supposedly this minor optimization has a dramatic effect on the execution of certain applications. If I remember correctly, Azure paid for/sponsored that work. Amazon may do the same when it sees statistics indicating there are enough people using FreeBSD on the platform to make it worth the cost of sponsoring such work. Problem is that FreeBSD users tend to be a bit more forward thinking and don't see value in cloud services (because... well, see the quote above for one reason ).

Another reason that might explain the little interest in using FreeBSD on AWS is that AWS was built around solving common issues that are really only present in Linux. Take for example the paradigm of treating your servers like cattle. Although it applies just fine for FreeBSD, it's targeted towards the Linux ecosystem where failures are better resolved by just trashing the instance and starting over.

 

[c2n]

It's pretty simple:
AWS has a leading market share running ephemeral instances, usually of Windows or Linux (probably 99%). It runs on a Linux framework. FreeBSD is not a priority, neither on the AWS side nor the FreeBSD side. FreeBSD's performance, in this case, is lower than expected, as observed.
Solution: Contact cloud vendor and ask for support. Contact FreeBSD mailing list (was done) asking politely for help and pointing out the issue. Issue bug report.

I agree with AWS being huge in the hosting industry and that FreeBSD needs to be on top of its game. I'm sure no one disagrees here, even if they're not invested in the industry.

FreeBSD has limited developers and limited funds. Seeing that it is not a problem elsewhere more investigation into the problem needs to be done. Maybe set up communication with FreeBSD devs and AWS and liaise at all levels.

On a side note: I don't agree with calling this forum useless nor claiming people have zero "fucks to give" or are in denial. It has nothing to do with being a grumpy old man but just plain rude. One reason why I and many others choose FreeBSD is the community, lets focus on the issue and not let emotions overrule.

 

[fraxamo]

You'll get help and support here but I don't think it's a direct line to key developers or movers & shakers in the FreeBSD world (I don't know where that is, though!)

There are a few developers that frequent these forums, but the majority will be found on the various specific mailing lists.

 

[Deleted member 66267]

A small cash donation entitles you to request one documentation fix.

A few milion dollars will help with bigger issues.

With this money I already have my own Enterprise Linux now and competing for the local market with other Enterprise Linux vendors