What if I never upgrade?

[Ole Juul]

Re: What if I never upgrade?

I have the same habit. For me it comes from running audio equipment where it is in fact better to leave it on. Electrolytics keep formed and stay healthy and that would be the same with computers except with modern switching supplies the caps are all small so I don't really think that matters. The real importance with the audio stuff (I'm showing my age) is the tubes don't like thermal stress. That, of course, has never been a computer problem since computers looked like anything most of us would recognize. Regarding thermal stress in LSIs, that sounds like BS to me. I don't think there is anything wrong with turning a computer on and off many times a day. However, like askrieger, I still don't.

My DOS machine doesn't care what you do to it software wise. You can turn the power off any time, and it only takes a few seconds to come back up to the state it was at. Linux machines not so. My personal Linux box is logged into other machines and after a normal 40 second boot up, it takes me many minutes to put it back together. I really dislike rebooting it and leave it up as long as I can. Even after updates it could take a few weeks before they will take effect because of that. Perhaps there is a solution to speeding up a Linux reboot and setup, but I've never spent a lot of time looking for one. The DOS machine boots and runs fast however, but that is why I run it. The idea is to have a machine where everything is as instant as paper and pencil. It is made specifically for recording and organizing my thoughts and if I had to wait, and then click on things and stuff like that, then it's raison d'être would be gone.

Regarding a UPS, I had one plugged in for a while, although just to keep it charged because I don't really need it. Around where I sit, there are literally piles of electronics and a lot of it is on. Last year I started noticing a high pitched squeal which became really annoying. I thought it was from a computer hard drive or fan although it was a peculiar high pitched sound which was really hard to pinpoint. So I went around banging everything to give something a jog and get it to shut up or identify itself. No go. Like an auto mechanic I then went around with a stethoscope to try to find the source. No go. After a lot of head scratching I then started to turn everything off - one by one. Until every single thing was off. The squeak was still there!!! I then went outside to see if there was some electrical device mounted anywhere. Nothing. Finally, while I was crawling around on my hands and knees trying to find the hidden bug.... there was the UPS. The battery had gone south and it was trying to tell me that it was low. Of course unplugging it didn't help because it was still low. No switch for that purpose. Finally I just put it outside and it squeaked for a week until it finally died. I hate those things.

[Snowhog]

Re: What if I never upgrade?

I have a laptop, so I don't keep it on all the time. I turn it on when I'm home, leave it on until I leave the house or go to bed, which ever comes first. My dad and his wife have desktops - he runs OS/2 and she Windows XP Pro - and they both keep their PCs on 24/7/365.

Thermal stresses are a real consideration. Even if such stresses are very small, they exist, and are cumulative. And over time, can result in breaks. Consider that powering up a PC induces the greatest stress on any component that can experience stress. While solid-state components likely experience the least stress, HDs experience the greatest when being spun up to speed.

As an analogy, starting a car creates the greatest 'wear' on the engine and it's adjunct components. Once the engine is running, if one keeps the oil and coolant clean, it will last much longer than the same engine that is repeatedly started and stopped.

[GreyGeek]

Re: What if I never upgrade?

Turning a computer on and off every day, or leaving it on all the time, depends a lot upon the quality of the hardware the computer is made, in addition to its operating environment. I have no problems turning a high quality laptop on or off as often as I want or is necessary. I cross my fingers when I turn cheap computers on or off because that is when current and voltage spikes occur most often. Voltage and/or current spikes can exceed component ratings, burning them out. It all depends on the safety factor chosen by the device maker. Better quality hardware also has current and voltage limiting circuity that prevents dangerous start up or shutdown currents or voltages. Cheap equipment rarely does.

BTW, improperly grounded electrical lines can bring those lightening induced current and voltage spikes into your equipment, on or off, and the results can be disastrous.

Universal Power Supplies (UPS) can give false hopes. When the power interrupts they experience the same spikes. The better ones, like better computers, have current and voltage limiting circuity. The poor ones fail. I've seen more than one server go down even though it had its own UPS, because the UPS failed when it was needed the most And, you don't test an UPS device by hitting its "test" button or yanking its plug out of the wall, unless you don't care that your server, and ALL the software it is running, has an abend. You down the server, then test the UPS. I used an ACP $250 500 Watt UPS for about three years. It was connected to my computer only, not my monitor nor the printer, because it didn't have the capacity to keep all of them running. It gave me about 5 minutes to power down my computer. Without a monitor one has to have the Ctl+Alt+Del key combo set to to a power down. I stopped using it when the test button failed. Besides, I found ReiserFS v3 so reliable in replaying its journal during power up that I had no qualms about a power out or yanking the plug out of the wall when thunder or lighting occurred outside.

Back to hardware. Take a resistor, for example. Say it is a 1/4 watt, 10K Ohms, +-20% when first made. Precision resistors can be +- 1% or better. When a resistor runs close to or at 1/4 watt of heat dissipation it can age much more quickly. The plasticizers and carbon binders can heat up and boil off, resulting in a change of resistance that affects the behavior of the circuit. Worse case mode is the resistor turning to a chunk of highly conductive Carbon, which lasts for only a few seconds before it burns through or sets fire to the circuit board! Depending on how it is made the resistance of the resistor can also increase with an increase in heat. Also, component placement can affect its survivability. Put that 1/4 watt resistor next to a 50 watt power transistor and the radiated heat absorbed by the resistor, along with its own heat, may exceed the rating of the resistor.

In a circuit the maximum current it can draw without exceeding its ability to dissipate heat is 5 ma, (P = I^2 * R), and the maximum voltage across that resistor cannot exceed 50V (P = I*E). Since the resistor is stock its resistance can vary between 8K and 12K Ohms, but its heat dissipation rating is still 1/4 watt. The max current will be between 5.6 ma and 4.6 ma. The max voltage will be between 44.6 volts and 54.3 volts.

A circuit designer has a choice to make, and costs determine what that choice will be. She can design a circuit in which the maximum voltage that resistor sees will be 25 v, a safety factor of two. Or, he can design his circuit so that the maximum voltage that resistor will see is 5 volts, a safety factor of 10. At 25v the resistor may have to dissipate 0.125 watts, but at 5v only 0.025 watts. Obviously, the cooler level will result in the resistor lasting a lot longer and maintaining its resistive value more closely. Combine a safety factor of 10 or more with a precision of 1% or more and you have military grade resistors, which is why military electronics, AND high end computers, work better and last longer. They cost more, too. Capacitors have to dissipate heat as well but the biggest problem with cheap capacitors is their seal. Their electrolytes can evaporate and the capacitor can change value or open entirely.

The switch to VLSI circuitry hasn't changed things, except to make circuits less expensive and more complex. Now, resistors are the size of pin heads and heat ratings are in milliwatts, but the same principles apply.

You get what you pay for. (Most of the time)

[Jean]

Re: What if I never upgrade?

I read this thread this morning and although I got a warning that this topic is old I still would like to reply.
I bought a new machine for a file/webserver/downloader. It's running xp now but I'd like to try some flavour of ubuntu. I have a spare machine I can test it on. I have tried xubuntu but it was just a bit too restricted so I went for 10.10.

I've seen kde 4 before and didn't like it very much. But I kept getting bad x crashes from x11vnc. I tried some other vnc server flavours but I did not find much useful. My feeling is the crashes are related to kde4. So I tried my a 8.04 setup and so I am back here again to tell what a relief it is to work in kde3.5 again. It faster, it looks nicer, it's more stable, I really do like it much more than kd4.

I've read about kde3.5 for meerkat I believe, but I fear that's going to be some niche thing.

My plan now is to go with 8.04 for now, if I don't get showstoppers the new machine is gonna run it too.

[GreyGeek]

Re: What if I never upgrade?

It's been my experience that KDE 4.3 onward has been solid as a rock. I am currently running Kubuntu 10.4 with KDE 4.5.3 and I don't have X server or desktop crashes. My Intel GM45 chip uses the Intel i915 driver.
Stability with X server and KDE depends more on your video chip than anything else, IMO.

After a period of time Intel, ATI and NVidia archive their "legacy" drivers, meaning that their current driver no longer supports the older chips. Users of older hardware are forced to install the legacy driver from the archive, but often they run into a catch-22: the older driver does not recognize the newer kernel. The user can't downgrade to an older kernel because many apps and other drivers depend on the capabilities of the newer kernel. Sometimes the solution is to use an older release, like 8.04 LTS, which has 3 year support on the desktop (4-21-2008 to 4-21-2011). After that, security and maintenance updates are no longer supported, projects stop being concerned about their compatibility with the out of date distro, and commercial vendors usually don't continue supplying apps for it. After March 21st of this year you will be own your own with 8.04. I'm not even sure the repository will remain up.

You mentioned KDE 3.5. The KDE desktop is made using the Qt API (Application Programming Interface), or toolkit. The KDE 3.5 DE was made using the Qt 3.x toolkit. Trolltech, which created Qt fifteen years ago, has ushered it through a series of upgrades. Each major change (1.x --> 2.x, 2.x --> 3.x and 3.x --> 4.x) experienced brief periods of controversy, generating the same kinds of complaints, except that the change from 3.x --> 4.x had a considerable amount of political gamesmanship involved as folks who didn't use KDE astroturfed forums where it was being discussed in order to write negative things about it.

Also, the change from KDE 3.5 to KDE 4.0 caused an event which didn't occur in the previous version changes, a "fork" by a group whose idea is to continue the support and development of KDE 3.x. It is a HUGE undertaking, and probably doomed to failure. Here is why. Trolltech had some decisions to make about Qt's development. The framework on which Qt 3 is based is relatively unchanged from the framework of prior releases. But, improvements in hardware (sound, graphics, acceleration, gesturing, IR sensors, fingerprint readers, etc..) that had taken place since the old Qt framework was developed were such that adding those capabilities to Qt 3 could be handled in only two way: adding code to the existing framework, or designing a newer, modular framework from the ground up. Just adding code would result in a "Rude Goldberg" API, much like just tacking rooms onto a house and trying to make them fit into the foundation, the plumbing, the heating and the electrical layouts. It would be much the same way Windows was developed, where all the 16 bit stuff was stuffed into the 8 bit framework, and had led to the problems with speed, stability and security that plagued Windows for a long time. Trolltech took the latter approach. They broke with the Qt 3 development model and designed Qt 4 from the ground up so that it was modularized, much like the Linux kernel was modularized and now one generally only has to use modprobe or rmmod to add or remove capabilities to the Linux kernel.

As one who was using Qt 3 for development I can tell you that development under Qt 4 was a breath of fresh air, and resulted in a vast improvement in productivity and capabilities. Much of the early KDE 4.0 to KDE 4.2.4 development was pure experimentation, where many ideas were floated, tried, and dropped, because it was easier to do than before, as developers plumbed the depths of the new API which, by the way, has continued to evolve. (Each new release of Qt 4.x has a change-log which tells what was added, changed or dropped from the previous version.). Some ideas were not as popular as others, but Qt 4 allowed KDE developers to do rapid changes to adjust to valid criticisms and suggestions. The result to date is KDE 4.6, which is an absolutely stunning desktop: beautiful, fast, powerful and flexible. The best on the planet, IMO.

I doubt that those forking KDE 3.5 will have enough years left in their lives to bring KDE 3.5 to the power of KDE 4.6, working on it part time. Here's why. On Feb 2nd, 2007 Trolltech announced that the "End Of Life" for Qt 3 would be July 1st, 2007 and said "At this time Qt 3 will reach end of life status and no further maintenance releases will be issued. " It's a matter of finances and resources. The difficulties are explained here. It's not a matter of "forcing" someone to use KDE 4, as many hotheads have claimed. KDE is free software and MOST of the people who work on it do so FREELY, on their own time and dime. THEY have chosen not to double their work and try to support the Qt 3 API *and* the KDE 3.5 desktop *and* keep KDE current by building on Qt 4. There is not enough hours in the day, or days in a year, or years in a life. The KDE 3.5.x Trinity Project was started by TWO guys. Now they list three contributors, counting themselves. Three. And, like most FOSS projects, they are begging for help and money. The spartan layout of their webpage demonstrates their lack of help and time. Qt 3 is a HUGE API. The Qt 3 has hundreds of Objects, each of which has hundreds more methods and properties. KDE 3.5 is a HUGE project. There are hundreds of applications written for KDE 3.5 using Qt 3.5.x. Many KDE 3.5.x apps have been archived as their developers have moved on to KDE 4.x.

That just KDE 3.5. EVEN IF those three guys could do the impossible and keep KDE 3.5.x current with new hardware, new drivers, new apps, and keep all of its existing utilities and 3rd party apps updated with security and bug fixes, the problem remains that each distro would have to devote a team of developers to integrate KDE 3.5.x with their existing distro base. That is not an easy job, and it takes MANY MORE than just 3 people. For the documentation on the next release of Kubuntu (Natty Narwhale) there are three guys working on just the documentation that you see when you click on the "help" button and they could use another dozen contributors if all of the topics are to be given more than a lick and a promise. The Trinity guys are at KDE 3.5.12. Unless they are near geniuses, independently wealth and can work full time plus evenings and weekends on KDE 3.5.x, burn-out will come sooner or later, and things with KDE 3.5 will be right back to where they were a year ago, when Trinity was started.

Your problems with KDE 4 might stem from a lack of learning how to run it. People weren't born with an innate knowledge of any particular desktop. I'd wager that if you reflect back to your early days of using KDE 3.5 you had learning frustrations over it as well. IOW, you became familiar with KDE 3.x because you "grew up" with it.

I doubt that the x11vnc crashes are related to KDE 4. You might try compiling it specifically for your machine, it you have the capability, but a better plan would be to install Kubuntu 10.4 (more than two years left on its LTS) and try some of the more mature VNC's available in the repository. One that has passed the 1.0 milestone. You can use the PPA to install KDE 4.5.3, which I find to be fast, stable and powerful.

[Jean]

Re: What if I never upgrade?

You have given me something to think about. Besides the history lessons.
Being a developer myself (I have used QT and know about the jump to 4), I understand about things having to grow and the need to redesign, as well as ever changing hardware.

In my case though, I don't have to live on a bleeding edge, it's not my main machine, it's just a file/webserver (and a bit of a play machine), which should stay up for a year or two after which I will evaluate again.

I thought about trying lucid thinking it might be more stable so now you mention it I am gonna try it, it's just the testmachine and I can put hardy back in 10 minutes with clonezilla if I want.

As for you saying that it takes time to learn a desktop, I will have to say that I just like the look and feel of kde 3 better than kde 4, no matter how more advanced it may be, but I am willing to try it (and fiddle with it over time to make it more to my liking) because I know it is more modern, if and only if I don't get showstoppers. And a desktop which restarts x every 5 minutes is not going to do it.

On the other hand, I will be wanting to use QT again for instance and it may not even work on hardy no more, still have to try that.

try some of the more mature VNC's available in the repository. One that has passed the 1.0 milestone.

Hm, my experience now and from some years ago with hardy, x11vnc is the best as a server. All others that I tried didn't work or had some other cons. The only one I could reliably run was krbf but it used too much cpu.

Thanks.

[lmilano]

Re: What if I never upgrade?

Jean: if what you need is a file and web server, and I assume the web-server will be accessed through the internet (as opposed to intranet), you need security.

My recommendation, given that you need both security and stability, and you don't need to upgrade all the time, is to use the latest ubuntu/kubuntu LTS release. That would be 10.04. More info here:
http://en.wikipedia.org/wiki/List_of_Ubuntu_releases

There is already an updated CD, which will have a lot of the updates in it (you should still run all the security updates on top of that):
http://releases.ubuntu.com/kubuntu/10.04.1/

For remote VNC type desktop, I honestly think you can't beat No Machine NX
http://kubuntuforums.net/forums/inde...opic=3105179.0

Everyone: you should not leave a computer connected to the internet running an unsupported operating system. The issue is the frigging security updates. If you do run an unsupported version, and someone runs a botnet in your machine, compromises your home network or breaks into the Pentagon from your machine and the FBI knocks at your door .... well, you'll probably regret being a lazy ass

[Jean]

Re: What if I never upgrade?

Feedback:

I installed lucid and it is more stable with x11vnc then 10.10 indeed. I had freenx working on lucid too but I prefer x11vnc because it connects much faster and I can have a shortcut for the connection on my xp desktop without always using the freenx viewer dialog.

While I still do like hardy's looks and feel, it just feels faster, I realize it has drawbacks to run such an 'old' version, for instance if I want to develop QT I'm better of with lucid.

Thanks.

[Ole Juul]

Re: What if I never upgrade?

Everyone: you should not leave a computer connected to the internet running an unsupported operating system. The issue is the frigging security updates. If you do run an unsupported version, and someone runs a botnet in your machine, compromises your home network or breaks into the Pentagon from your machine and the FBI knocks at your door .... well, you'll probably regret being a lazy ass

I run an number of "unsupported" OSs and I resent being called a lazy ass. I also do not have a security problem. Thanks for the heads up, but there is a little more to it.

[askrieger]

Re: What if I never upgrade?

Personally, I don't think you're a lazy ass. Anyone looking at your posts here would know that is not the case. However, I do think that you may (or may not) have a security problem. Have you looked at your /var/log/auth.log file lately? I posted my eye-opening experience HERE.

[lmilano]

Re: What if I never upgrade?

Yes, sorry, that meant to be a humorous comment (hence the smiley). And you, in particular, Ole, are great, I always see you helping others.

On the topic of security, though, I can't be wishy washy. If I am to run something behind doors (not connected to the net), I can do whatever I please. But, if I am connected to the net, I need to be responsible about it. Someone can compromise _my_ machine, but that can affect _other_ users. This is like driving a car without brakes, and saying "hey, this is my car, what do others care".

I know, millions, perhaps hundreds of millions of people drive on the net without breaks. So, we need to open their eyes. Askrieger, great post, thanks for the link!
Cheers!

[lmilano]

Re: What if I never upgrade?

Feedback:

I installed lucid and it is more stable with x11vnc then 10.10 indeed.

Awesome, I think this is best for your needs, thanks for touching base!

[Ole Juul]

Re: What if I never upgrade?

Yes, sorry, that meant to be a humorous comment (hence the smiley). And you, in particular, Ole, are great, I always see you helping others. ...

Actually, my apologies. I should have expressed myself differently. No offence taken.

. . . Have you looked at your /var/log/auth.log file lately? I posted my eye-opening experience HERE.

I just looked at that and it had nothing but iterations (with date changes) of this:

Feb 10 09:39:01 SCO CRON[15519]: pam_unix(cron:session): session opened for user root by (uid=0)

I think that being behind several levels of NAT makes a difference. I don't know what is more common, being straight on the net with a public IP or like me, not really having a public IP. To me my situation is normal because that's what "everybody's" got. I wouldn't want to bother running servers at home anyway, since commercial server space is dirt cheap nowadays. (I pay $6/mo for 750GB traffic)

Actually, I guess Kubuntu 8.04 is going to be unsupported shortly. Since that's what I'm running, I guess I'll have to make some kind of decision soon. However, I was really thinking about DOS in my case. I have a DOS box on the net 24/7. No one is adding any security holes so it remains impenetrable, outside of social engineering or physical (hammer) attack. Am I being cocky? Perhaps, but I can't actually imagine an attack vector and I've never heard of one.

I would think it also depends on what you are using your computer for. Perhaps we all assume that there is Firefox or other major browser running, and that's fair enough because it is usually the case. However, it is quite possible to just use SSH and FTP and other cools stuff like that on the net. A box for just network analysis is also a definite reality. I usually have a laptop available with some version of Linux but no GUI. How vulnerable is that going to be if I don't update it?