Sorry. Can't answer that one. I run my own mail server so I wanted reliability. I was strictly interested in using RAID to nmake the box so it would still boot even if a drive died.
With regard to your question:
Pros: HPT-37X is cheap
Cons: Striping takes almost no CPU time since its just a matter of figuring out which drive has the particular piece of data if you use Linux software RAID.
I've had flawless Linux software RAID 0 performance on another box but its a dualy Athlon 2400+ so it has more than enough spare CPU cycles to handle software RAID.
The SCO-math is probably what got Darl and company all exited. Think about all those thousands if not millions of Linux based consumer products and SCO thinks they can extort a licensing fee for using Linux on them. If they were to win their case, Darl and company would probably end up with more money than Bill Gates for doing nothing but filing a few lawsuits. Liability lawyers have been known to take up cases on a contingency basis with a lot less chance of winning, not as much money at stake and less legal basis. The odds are still better than the lottery.
There's a reason I keep calling SCO management "a bunch of ex-ambulance chasers." They've graduated to the big time.
[dave@fraud: ~]#/tmp/hpt > file hpt37x2lib.o hpt37x2lib.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped
for which no source is provided. Guess what happens if you try to use it with an updated kernel? But thanks for confusing everyone about whether there is really an open source driver for the HPT controllers. BTW, you don't need hpt37x21lib.o if you aren't using RAID according to the readme.txt (which is what most people recommend).
What should it have done? The reason linux can boot up it situations like that with the software RAID is because it isn't real RAID. You are mirroring the disk through software -- that isn't RAID.
Suggest you read up on why people use RAID for other than RAID 0. The idea is *reliability*. You use RAID if you want a system that keeps running even if a drive craps out. The system in question is on a UPS and uses ext3 for all file systems mounted "auto" through fstab. Also look into how a real RAID controller (e.g., like the various 3ware products) handles this situation.
Linux supported only with binary drivers so you are stuck at whatever version of the kernel Highpoint decides to support.
Mine decided that the two drives I had set up in RAID 1 were out of sync so EVERY time I booted it stopped during its BIOS check and asked me what I wanted to do about it. I wanted a RAID controller that would keep my system going even if one of the mirrored drives bit the big one. Bzzt. Wrong answer.
I decided to see how it handled a drive failure in case I found a work around for the above problem. I powered down and removed the IDE ribbon cable from one of the mirrored drives and then powered up. Same approach as above. Stupid BIOS says that one of the drives is down and hangs until I tell it to come up anyway. Bzzt. Way wrong answer.
There is no way to tell the Highpoint BIOS not to do this but to boot as best it can with the surviving hardware it finds.
The only thing I ended up using mine for was to configure the drives I wanted mirrored as the boot device. I re-ran the above tests after setting up the drives as Linux software RAID 1 and it performed flawlessly. Drives out of sync: a little message in dmesg and/var/log/messages and they get synced in the background after the system is up. Pull an IDE cable off of either drive and you get the same result. System boots and there's a message saying the RAID is in degraded mode.
So my home server is now set up such that I can lose one drive from either mirror and it will still be fully functional. I could probably set the BIOS boot order such that I don't need the HPT370 at all to make this happen but its on the board and I may as well get at least some use out of it. At least I didn't pay that much more for a motherboard with built0-in RAID but I sure wasted enough time trying to get the thing to give me what I wanted from a RAID controller: reliability through redundancy.
There are already two of them. Interestingly, the first one has been modded down to -1 while the second has been modded up to +4. I guess some moderators are as humor challenged as you.
There is also the obligatory "In Soviet Russia" comment as well as expected SCO licensing joke. So far, the only thing that's missing is a CowboyNeal joke but those generally only show up on polls.
But I can make up something really lame concerning CowboyNeal and super computers if you like...
Unfortunately, it looks more like Darl wants to do an episode of "Trading Spaces" with Linux. Something like he gets to bring over all the good Linux ideas into SCO in exchange for licensing his vaunted SCO Unix Intellectual Property.
This makes for an interesting approach to covering all the Linux IP that SCO has been stealing.
Rusty made changes to module-init-tools (modprobe, lsmod, insmod and friends). You need to install an updated version of module-init-tools in order for any kernel after about 2.5.56 to correctly load modules. Alternatively, you can compile a monlithic kernel.
In order for some idea to be patentable. it most have a unique, functional, tangible expression and it must truly be innovative (not otherwise exist).
I think you'll find that this nicely leaves out patent abominations such as software and business practices. Software is sufficiently protected by copyright. Business practices have not been subject of intellectual property law until recently. Including them under any form of IP is a mistake.
I disagree with your comment about trade secrets and NDAs. Once something has a tangible expression and is being sold, it is difficult to impossible to keep a competitor from disecting it. Patents are effective at preventing someone else from simply imitating. They at least have to figure out how it was done and then how to do it differently enough to not infringe.
I'm a working stiff. I put in my 40 hours plus at my day job and I run a small internet site over an ISDN line to keep up my current skills and learn new ones so I stay employed. This keeps me rather busy plus I've been running a development Linux kernel since the low 2.5.20s because I also want to learn about the kernel. Somehwere out of all the things I've been doing I could probably come up with material for a book but it would probably mean investigating things I'm not really all that interested in or that aren't necessary for what I'm doing in order to make the work complete. It would also require even more of my so called "leisure time"
Get real. I'm not excessively greedy but I am human and I live in the real world. This wouldn't be a work of art since were talking about a technology how-to that has a limited life span so don't give me "art for art's sake" bull. As for the "reputation" incentive, a prof of mine once said before leaving the university for the corporate world, "You can't eat prestige or reputation." Finally, how do I justify to my wife that I need to spend even more time messing with computers and not doing stuff around the house unless I can at least point to some possible monetary reward from my effort?
So I'm not starving but, damn I'm tired at the end of the week. I need a tangible incentive (hint: $$$$) to work harder. When you have such an incentive other than some sort of reasonable renumeration based on copyright, let me know. Otherwise, I need more sleep more than I need to enhance my reputation.
So, how many GHz is that corn stalk? Oh, it doesn't.
So, what diseases does your corn stalk cure? Oh, it doesn't.
So, how many people are entertained by watching your corn stalk grow? Oh, not many.
So, how many people does your corn stalk seat and how fast does it do 0 to 60? Oh, it doesn't.
So, what diminished costs do the corn pirates face if they buy some of my corn and plant it? No cost of owning the farm to grow it on? No fertilizer? No harvesting costs? No silage costs? Oh, they're the same unless they're either more efficient or come up with a better process for growing corn. In fact, if they're as stupid as your analogy, they'll probably assume it plants itself and harvests itself in thin air.
Get a grip man. At least attack my ideas with something that's a relevant comeback.
The concept of intellectual property was created so that people and companies who invest in the creation of new "things" could re-coup their development investment. This is true for writers, artists, inventors, R&D departments, etc. Some, such as trademark laws, were created to protect consumers from unscupulous people providing fraudulent imitations of recognized products (i.e., they are *VERY GOOD* for consumers).
Patents and copyrights were intended to provide an incentive for people to create new things. As an example, if I am an author, what is my incentive to continue to write if my works can be freely copied? Likewise, why should a pharmaceutical comapany work to discover, refine and test a new medecine if the moment it comes out anyone else can make their own copy of it without incurring the development costs?
Intellectual property laws are a necessity for modern society. Sadly, some people like SCO and PanIP have subverted those laws to try to gain from works they had nothing to do with. Luckily for the open source community, the ambulance chasers at SCO were stupid enough to go after somebody big instead of being bottom feeders like PanIP and just hitting little guys for licensing fees.
... and here I was hoping that threatening them with barratry was like threatening them with some form of medieval torture like garroting or bastinado. ESR is being way too nice.
SCO has been doing their best to get their legal disputed tried in every venue except a court of law where they know they'll loose. If ESR had simply said, "I disagree with Darl McBride and SCO," there would be no press coverage whatsoever. What he has done, instead, is to fire off a equal but opposite inflamatory rant to match the spewage from Lindon, UT. There is no threat contained in his rant other than that SCO will wish their ex-ambulance chaser management team hadn't provoked the open source community. More precisely, there is no threat of either physical violence or illegal action and thus, nothing illegal.
I remember reading someplace that Linus wants to have point releases for the stable kernel every couple of months now. No more of these massive releases that take forever to come out like 2.4.19 or 2.4.21. Another change has been to keep the current stable kernel "open" to patches so that kernel.org has had both the 2.4.21 kernel (as released) and a bitkeeper patch snapshot such as 2.4.21-bk38. Don't be surprised if there is a 2.4.22-bk1 within a day or so.
The local papers here in Denver have all been complaining because the Princeton Review claims that the University of Colorado is the #1 party school. Here's a link to an article carried by "The Guardian" or another in the Rocky Mountain News. One thing nice about partying at a mile above sea level is there's already less oxygen so you can achieve the same buz on less total consumption for those of you on a budget.
This is the same strategy the same people used when they sued Microsoft over DRDOS as Caldera. That time it worked. A whole bunch of people all saw little, tiny Caldera taking on big, bad M$ and winning and thought it was "a good thing."
Now the same people are once again posturing as a "poor little upstart company" taking on the big, bad established company. And to make matters worse, that big, bad company isn't fighting fairly and is manipulating the world into joining in their attack of gallant little SCO. This is the same tactic as when a sleaze-ball lawyer has their fake injury client show up in court wearing a neck brace that they put on in the car and take off as soon as the jury is out of sight. Get to the jury on sympathy when your facts don't support your case. Of course, SCO management is made up of a bunch of sleaze-ball lawyers so why should we expect any better?
To some extent, the same thing can be said about their attacks on OSS. Only with Open Source the claim is that we're a bunch of communists out to destroy capitalism. They want to manipulate this into them being the "good guys" and everyone they're attacking (IBM, Open Source community, Linux users) are the "bad guys."
OK. Connecting the teletype to the internet probably violates one of SCO's copyrights. In fact, I'd really believe this since the teletype and their code base are about the same vintage.
I agree with the guy. There are three SCO stories on the front page right now. Do we really need to debate SCO's every (rather predictable) move? This is worse than the days when every other story was a dupe.
So you would rather be surprised one day to find a bill from SCO in your mailbox demanding that you pay them $699.00 (plus accrued interest) for using Linux since the time they asserted their ownership of the product? And this time the bill is legit because some obscure judge decided they really did own "UNIX" and anything that looked like it in a default judgement because no one cared enough to contest the claim. Oh, and there's a guy from Mario's Collection Agency who would be happy to explain to you why your kneecaps were worth more to you than the $699.00 (plus accrued interest) that SCO said they were entitled to.
Bad laws get made and bad ruling are handed down even when the light of public interest is on but this happens much more frequently when people stick their head in the sand and say they don't want to hear about it. Grow up and pay attention. Sometimes important stuff ("...stuff that matters") isn't about some new, whiz-bang gadget.
At $350K to upgrade, you are talking a serious number of systems. I'm not saying you can talk them into giving it to you for free but whoever is doing the purchasing should be able to negotiate something better than full retail. You are mainly buying support so things to point out include multiple identical systems, internal support for end-user systems, etc. that mean they won't have to answer too many really dumb questions.
Favorite really dumb support question: do I have to plug it into the electricity?
Slashdot Mirror
Sorry. Can't answer that one. I run my own mail server so I wanted reliability. I was strictly interested in using RAID to nmake the box so it would still boot even if a drive died.
With regard to your question:
Pros: HPT-37X is cheap
Cons: Striping takes almost no CPU time since its just a matter of figuring out which drive has the particular piece of data if you use Linux software RAID.
I've had flawless Linux software RAID 0 performance on another box but its a dualy Athlon 2400+ so it has more than enough spare CPU cycles to handle software RAID.
The SCO-math is probably what got Darl and company all exited. Think about all those thousands if not millions of Linux based consumer products and SCO thinks they can extort a licensing fee for using Linux on them. If they were to win their case, Darl and company would probably end up with more money than Bill Gates for doing nothing but filing a few lawsuits. Liability lawyers have been known to take up cases on a contingency basis with a lot less chance of winning, not as much money at stake and less legal basis. The odds are still better than the lottery.
There's a reason I keep calling SCO management "a bunch of ex-ambulance chasers." They've graduated to the big time.
for which no source is provided. Guess what happens if you try to use it with an updated kernel? But thanks for confusing everyone about whether there is really an open source driver for the HPT controllers. BTW, you don't need hpt37x21lib.o if you aren't using RAID according to the readme.txt (which is what most people recommend).
Suggest you read up on why people use RAID for other than RAID 0. The idea is *reliability*. You use RAID if you want a system that keeps running even if a drive craps out. The system in question is on a UPS and uses ext3 for all file systems mounted "auto" through fstab. Also look into how a real RAID controller (e.g., like the various 3ware products) handles this situation.
Thought about doing a simple
$
but figured too many wouldn't understand. The other is my standard *user* prompt on my home box. That's why it says dave@fraud.
Diplomacy is the art of telling someone to "Go to Hell" in such a way that they think they'll enjoy the trip.
--Caskie Stinnett
Your original quote is ("...find a big stick.") is attributed to Wynn Catlin also speaking about diplomacy.
Short version: it sucks.
/var/log/messages and they get synced in the background after the system is up. Pull an IDE cable off of either drive and you get the same result. System boots and there's a message saying the RAID is in degraded mode.
Long version:
Linux supported only with binary drivers so you are stuck at whatever version of the kernel Highpoint decides to support.
Mine decided that the two drives I had set up in RAID 1 were out of sync so EVERY time I booted it stopped during its BIOS check and asked me what I wanted to do about it. I wanted a RAID controller that would keep my system going even if one of the mirrored drives bit the big one. Bzzt. Wrong answer.
I decided to see how it handled a drive failure in case I found a work around for the above problem. I powered down and removed the IDE ribbon cable from one of the mirrored drives and then powered up. Same approach as above. Stupid BIOS says that one of the drives is down and hangs until I tell it to come up anyway. Bzzt. Way wrong answer.
There is no way to tell the Highpoint BIOS not to do this but to boot as best it can with the surviving hardware it finds.
The only thing I ended up using mine for was to configure the drives I wanted mirrored as the boot device. I re-ran the above tests after setting up the drives as Linux software RAID 1 and it performed flawlessly. Drives out of sync: a little message in dmesg and
So my home server is now set up such that I can lose one drive from either mirror and it will still be fully functional. I could probably set the BIOS boot order such that I don't need the HPT370 at all to make this happen but its on the board and I may as well get at least some use out of it. At least I didn't pay that much more for a motherboard with built0-in RAID but I sure wasted enough time trying to get the thing to give me what I wanted from a RAID controller: reliability through redundancy.
I don't see what all the argument is about. There is a standard *nix desktop. It looks like:
#
or you can customize like:
[dave@fraud ~]#
There are already two of them. Interestingly, the first one has been modded down to -1 while the second has been modded up to +4. I guess some moderators are as humor challenged as you.
There is also the obligatory "In Soviet Russia" comment as well as expected SCO licensing joke. So far, the only thing that's missing is a CowboyNeal joke but those generally only show up on polls.
But I can make up something really lame concerning CowboyNeal and super computers if you like...
Unfortunately, it looks more like Darl wants to do an episode of "Trading Spaces" with Linux. Something like he gets to bring over all the good Linux ideas into SCO in exchange for licensing his vaunted SCO Unix Intellectual Property.
This makes for an interesting approach to covering all the Linux IP that SCO has been stealing.
Stable as a rock:
[dave@bend ~]# uname -a
Linux bend.local.davenjudy.org 2.6.0-test4 #1 SMP Sat Aug 23 10:15:04 MDT 2003 i686 athlon i386 GNU/Linux
[dave@bend ~]# uptime
10:31:57 up 3 days, 22:02, 9 users, load average: 0.06, 0.01, 0.00
and much more responsive than 2.4. I'm at work right now so this system is basically idle but it gets hammered when I'm at home.
You can get the updated module-init-tools at:
ftp.kernel.org/pub/linux/kernel/people/rusty/modul es
The latest is module-init-tools-0.9.14-pre1.tar.bz2
Somewhat shorter:
In order for some idea to be patentable. it most have a unique, functional, tangible expression and it must truly be innovative (not otherwise exist).
I think you'll find that this nicely leaves out patent abominations such as software and business practices. Software is sufficiently protected by copyright. Business practices have not been subject of intellectual property law until recently. Including them under any form of IP is a mistake.
I disagree with your comment about trade secrets and NDAs. Once something has a tangible expression and is being sold, it is difficult to impossible to keep a competitor from disecting it. Patents are effective at preventing someone else from simply imitating. They at least have to figure out how it was done and then how to do it differently enough to not infringe.
I'm a working stiff. I put in my 40 hours plus at my day job and I run a small internet site over an ISDN line to keep up my current skills and learn new ones so I stay employed. This keeps me rather busy plus I've been running a development Linux kernel since the low 2.5.20s because I also want to learn about the kernel. Somehwere out of all the things I've been doing I could probably come up with material for a book but it would probably mean investigating things I'm not really all that interested in or that aren't necessary for what I'm doing in order to make the work complete. It would also require even more of my so called "leisure time"
Get real. I'm not excessively greedy but I am human and I live in the real world. This wouldn't be a work of art since were talking about a technology how-to that has a limited life span so don't give me "art for art's sake" bull. As for the "reputation" incentive, a prof of mine once said before leaving the university for the corporate world, "You can't eat prestige or reputation." Finally, how do I justify to my wife that I need to spend even more time messing with computers and not doing stuff around the house unless I can at least point to some possible monetary reward from my effort?
So I'm not starving but, damn I'm tired at the end of the week. I need a tangible incentive (hint: $$$$) to work harder. When you have such an incentive other than some sort of reasonable renumeration based on copyright, let me know. Otherwise, I need more sleep more than I need to enhance my reputation.
So, how many GHz is that corn stalk? Oh, it doesn't.
So, what diseases does your corn stalk cure? Oh, it doesn't.
So, how many people are entertained by watching your corn stalk grow? Oh, not many.
So, how many people does your corn stalk seat and how fast does it do 0 to 60? Oh, it doesn't.
So, what diminished costs do the corn pirates face if they buy some of my corn and plant it? No cost of owning the farm to grow it on? No fertilizer? No harvesting costs? No silage costs? Oh, they're the same unless they're either more efficient or come up with a better process for growing corn. In fact, if they're as stupid as your analogy, they'll probably assume it plants itself and harvests itself in thin air.
Get a grip man. At least attack my ideas with something that's a relevant comeback.
The concept of intellectual property was created so that people and companies who invest in the creation of new "things" could re-coup their development investment. This is true for writers, artists, inventors, R&D departments, etc. Some, such as trademark laws, were created to protect consumers from unscupulous people providing fraudulent imitations of recognized products (i.e., they are *VERY GOOD* for consumers).
Patents and copyrights were intended to provide an incentive for people to create new things. As an example, if I am an author, what is my incentive to continue to write if my works can be freely copied? Likewise, why should a pharmaceutical comapany work to discover, refine and test a new medecine if the moment it comes out anyone else can make their own copy of it without incurring the development costs?
Intellectual property laws are a necessity for modern society. Sadly, some people like SCO and PanIP have subverted those laws to try to gain from works they had nothing to do with. Luckily for the open source community, the ambulance chasers at SCO were stupid enough to go after somebody big instead of being bottom feeders like PanIP and just hitting little guys for licensing fees.
... and here I was hoping that threatening them with barratry was like threatening them with some form of medieval torture like garroting or bastinado. ESR is being way too nice.
SCO has been doing their best to get their legal disputed tried in every venue except a court of law where they know they'll loose. If ESR had simply said, "I disagree with Darl McBride and SCO," there would be no press coverage whatsoever. What he has done, instead, is to fire off a equal but opposite inflamatory rant to match the spewage from Lindon, UT. There is no threat contained in his rant other than that SCO will wish their ex-ambulance chaser management team hadn't provoked the open source community. More precisely, there is no threat of either physical violence or illegal action and thus, nothing illegal.
It is an excellent rant though.
But it doesn't sound like someone burning rubber on asphalt... which is the sound effect Hollyweird attaches.
The first rule when viewing anything from Hollyweird is:
:-)
==> Suspend Disbelief ==
Examples:
Count how many times someone's six-shooter shoots without being reloaded in almost any classic western or detective/police movie.
Tires squealing on gravel roads.
Male lead is a geek but doesn't look like a nerd in "Blowfish".
I remember reading someplace that Linus wants to have point releases for the stable kernel every couple of months now. No more of these massive releases that take forever to come out like 2.4.19 or 2.4.21. Another change has been to keep the current stable kernel "open" to patches so that kernel.org has had both the 2.4.21 kernel (as released) and a bitkeeper patch snapshot such as 2.4.21-bk38. Don't be surprised if there is a 2.4.22-bk1 within a day or so.
The local papers here in Denver have all been complaining because the Princeton Review claims that the University of Colorado is the #1 party school. Here's a link to an article carried by "The Guardian" or another in the Rocky Mountain News. One thing nice about partying at a mile above sea level is there's already less oxygen so you can achieve the same buz on less total consumption for those of you on a budget.
This is the same strategy the same people used when they sued Microsoft over DRDOS as Caldera. That time it worked. A whole bunch of people all saw little, tiny Caldera taking on big, bad M$ and winning and thought it was "a good thing."
Now the same people are once again posturing as a "poor little upstart company" taking on the big, bad established company. And to make matters worse, that big, bad company isn't fighting fairly and is manipulating the world into joining in their attack of gallant little SCO. This is the same tactic as when a sleaze-ball lawyer has their fake injury client show up in court wearing a neck brace that they put on in the car and take off as soon as the jury is out of sight. Get to the jury on sympathy when your facts don't support your case. Of course, SCO management is made up of a bunch of sleaze-ball lawyers so why should we expect any better?
To some extent, the same thing can be said about their attacks on OSS. Only with Open Source the claim is that we're a bunch of communists out to destroy capitalism. They want to manipulate this into them being the "good guys" and everyone they're attacking (IBM, Open Source community, Linux users) are the "bad guys."
OK. Connecting the teletype to the internet probably violates one of SCO's copyrights. In fact, I'd really believe this since the teletype and their code base are about the same vintage.
So you would rather be surprised one day to find a bill from SCO in your mailbox demanding that you pay them $699.00 (plus accrued interest) for using Linux since the time they asserted their ownership of the product? And this time the bill is legit because some obscure judge decided they really did own "UNIX" and anything that looked like it in a default judgement because no one cared enough to contest the claim. Oh, and there's a guy from Mario's Collection Agency who would be happy to explain to you why your kneecaps were worth more to you than the $699.00 (plus accrued interest) that SCO said they were entitled to.
Bad laws get made and bad ruling are handed down even when the light of public interest is on but this happens much more frequently when people stick their head in the sand and say they don't want to hear about it. Grow up and pay attention. Sometimes important stuff ("...stuff that matters") isn't about some new, whiz-bang gadget.
At $350K to upgrade, you are talking a serious number of systems. I'm not saying you can talk them into giving it to you for free but whoever is doing the purchasing should be able to negotiate something better than full retail. You are mainly buying support so things to point out include multiple identical systems, internal support for end-user systems, etc. that mean they won't have to answer too many really dumb questions.
Favorite really dumb support question: do I have to plug it into the electricity?