Score:5, Informative? (No offence Michael) People who moderate Informative are by default ignorant, or else it would'nt be informative to them. So of course, they are not qualified to say this is "Informative" without your post having supporting evidence, links, etc.
The thought that it's "not the drives but bad handling" is silly. There are bad and good batches. It seems from this story that there might be a bad batch of IBM's. Do you think that mostly these IBM hdd owners for some unknown reason are handling or treating their drives baddly whereas the other owners are more enlightened on drive care? Not a chance.
Poor PSU quality, heat removal, shock, vibration and simply bad manufacturing batches or even poor design (rare to get past Q-Ctrl before mass mfg) can be the causes of drive failure. The problem CAN be the drives and in this case seems to be.
The damage done by a bump is usually not evident for months until the drive fails. Apparently a drive will get a small mechanical irregularity, and then slowly chew on itself until failure.
An impact strong enough to cause the hovering heads to actually touch the moving platters can cause the paint like magnetic surface to fleck, causing bad blocks. In time, as the heads pass over that area, they can cause further chipping off of the surface and thus the bad blocks grow in numbers. This may result in data loss and possibly destruction of heads alltogether.
With modern drives, that automatically re-map defective blocks to spares set aside to guarantee maximum usable blocks, this problem can be masked until those spare blocks are exhausted.
You need a high enough impact or vibration to cause a head to come down onto a spinning platter.
If you put a tower case on the floor, and the floor moves a little every time someone walks near, expect problems.
Vibration levels of a person walking near a PC on the floor is nowhere near high enough to cause any problem. The vibration induced by the drives own head movement will cause far quicker acceleration and total drive movement than that (which is mostly a safer horizontal movement). Unless the PC is on creaky floorboards, but even then this is *slow* movement.
If you put a case on a concrete floor, but it is often knocked during the day, expect failure.
Yeah, if the PC itself is knocked, it could cause drive problems, depending on what knocked it (how hard, heavy and fast the object was moving). But I imagine people who care for thier PC's actually take care of them.
If a computer is on a table that moves a little while you are working, it may not last long.
The table is bound to take a lot of the usual day to day shock, which should be quite minimal. Sure if you're dropping heavy objects onto your desk with your PC on it then you might be at risk. But not typing, mousing, paper folders, general office duties type stuff or even dropping a book onto your table. The table stops the book, not the hdd, and any vibration that gets to the hdd would be very damped by the desk.
This failure mode is dependent on how much movement about the axis actually happens, of course.
Which axis? Drives don't like to be yawed while they are spinning as the gyroscopic effect puts strains on the bearings in ways they were'nt designed for. Sideways shock should cause minimal problems, it is the vertical shock that causes heads to crash.
Drives are built to handle a lot of Gs when they are not powered, but when they are running they are very vulnerable.
Very true. Usually 4 or 5 times. Drives are also more susceptible to vibration than shock. My drives takes 300Gs shock non-operating, 63 operating, but only 0.5Gs vibration while operating!
Put a drive on its own power supply connector.
Every PSU I have ever had cause to open, had all it's GND, +5V and +12V wired to common power rails within the PSU. Seperate connectors barely make a difference. Anyway, drives tend to not use much power once they have reached their stated spindle speeds after power up. My drive draws 27 Watts during power-on, then 11.6 during seeks, 8.5 reading/writing and only 7 Watts during idle (platter spinning, no head movement or RW). This is a 7200 rpm drive and I have two of them.
I guess one possible saving grace at boot time is that the CPU is not doing a real lot while the hdds are spinning up. As an interesting comparison, my Intel Pentium II 300MHz, draws 43 Watts at peak!
I've had good luck with considerable quantities of Western Digital drives. Good support, also. I've had bad luck with Quantum, Seagate, and Maxtor.
Problem is, that ALL manufacturers have particularly bad batches and good batches. I've personally had good experience with WD, IBM and Seagate, so-so Quantum and absolutely dismal experience with Maxtor (I will never buy one again). WD is known to have past models that do NOT adhere to standards that the drive reports to support.
Drive manufacturers in the past (not sure if they still do), sold "special" models that passed as a grade A drive, built to higher tolerances (almost as if mil-spec) and sold for higher prices for server usage. A DEC rep told me this years ago and looking at the lifetimes and warrantee periods of the high end SCSI drives we were using, I would tend to beleive him.
Recently, while trying to find out why Linux 2.4.x software RAID-0 device performance was about 18% SLOWER than the/dev/hd# devices that make up the stripe, I decided to do a simple little test from single user mode that I thought might show the maximum read speed I could expect from my Linux kernel (2.4.10 at the moment, as tested).
So I did a (probably very unscientific) # dd if=/dev/zero of=/dev/null with a count= of 1Gb, timed it and worked out the transfer rate... ~34M bytes/sec. WTF?!?!
I then did the same in FreeBSD 4.4 Release.................... 250M bytes/sec!
I'm now off to wipe some adjacent/dev/hda and/dev/hdc partitions to set up FreeBSD software RAID-0 to see if I should move my desktop #1 from Linux to FreeBSD.
PS, I have a PII-300 (100MHz FSB) with 256Mb PC100 SDRAM for comparison.
Troll...Do you live in...how about telling the story as it really is...just remember the damn thig was there 20 years before anyone esle...be in Sydney...70 k away...You or your parents...
What a great big steaming, anonymous, assuming pile of horse shit.
I have lived in Bondi for almost 30 years. My mother and her father also at a nearby beach for much longer than that.
Lucas Heights is about 28km from the Sydney CBD.
I prefer that a "fairly safe reactor" be way out past Dubbo than 28km from SYDNEY CBD. Especially when the new proposed site docco states "the cut-off dose is reached 25 km from the proposed site", they also speak of doses out to 50km in an accident situation being well under safe levels. Hey I've got an idea, why don't they just put it 500km inland!?
Are the residents of the Leichhardt, Ashfield, Strathfield, Fairfield, Bankstown, Marrickville, Canterbury, Rockdale, Hurstville and Kogarah areas to be blamed for living within the 25km unsafe zone also?
Oh man, it is rare for me to enjoy a hollywood flick. They usually have some bullshit device, ability or random buzzwords to help patch up an awful script. Which is especially annoying when you recognise the buzzwords or see that parts of the movie are extremely over explained with that shit for the stupid audience.
I know they're often meant to be amazing to watch but fuck. Swordfish, Broken Arrow, Face Off, Anaconda, MI2, etc etc etc.
I enjoyed Alien, The Matrix, The Sixth Sense, Shrek and a few others I can't remember right now, but most of what I see I think I could get more entertainment out of masturbating with a handfull of rusty razor blades.
I like sci-fi, but I don't when I am treated like an idiot with over simplified bullshit. If they have to describe something non-existant, at least make up a new word for it! Instead of "ip-chains multiplexor" for a time machine, or some shit like that.
Anyone know of any web sites that quote the bullshit from movies?
Plus, imagine if those planes were to be ploughed into reactors around the US.
I live in Sydney, we have this tiny little reactor that is for medical use. Regardless I'm not real pleased with the fact that IT WAS a target during the games and probably will be forever more, until a sane government decides to shut it down and relocate to a safer area away from innocents.
I suppose you'll get used to that lightness after while but it just feels odd.
I used to like the original MS mouse for it's weight, but now I have an optical Logitech wheel mouse, when ever I use someone elses mouse I find it feels too heavy and usually jumpy.
Although the high speed tracking problem can cause problems at times, I find nothing as smooth (both physical feel and tracking) and love the light weight.
A mate and I set up a Quake2 party complete with a digital projector set up on a seperate machine on the network using the "camera where the action is" mode and..... no bloody one came. Everyone wanted to go see anyother stupid hollywood flick.:(
Re:Does PCS *need* GPS for positional data?
on
GPS Meets PCS
·
· Score: 1
I really, really, really doubt this.
I remember reading that the GSM StarTAC showed within it's secret network menu, Mobile-Base distance in metres. However, in the interest of correct info and to the detriment of my karma, I delved into the GSM specs and found...
That the value that can be used to calculate distance is a 6 bit number that divides up the max distance within the 35.2km usable time window, giving an accuracy from one sample (one cell) of about 550 metres.
So your value of 400m for multiple cells is probably correct. One thousand appologies.
Does PCS *need* GPS for positional data?
on
GPS Meets PCS
·
· Score: 3, Interesting
In Australia, with GSM, you call 000 or the international GSM emergency number (I forgot it), and it will automatically call the Fire/Ambo/Police number through your network provider (Telstra, Optus, Vodaphone or a reseller), if you have no signal to your own provider it will allow usage of any provider you have signal with for your emergency call. Even without a SIM card inserted in the phone. This is a legal requirement and seems to be a feature built in to GSM itself.
Due to the very precise time division multiplexing used with GSM, the distance you are from the base station you are currently subscribed can be gleaned down to a metre. If they can force your phone to switch to 2 other cells after an emergency call, they could probably pin point you without GPS. With the hidden Network menus in Motorolla StarTac GSM and Nokia phones, you can see how far you are from the base station in metres.
there are other reasons why many still consider Linux a toy.
I've been using Linux for about 4 years, and advocate it based on the fact that it rarely (once or twice) crashes on me (the kernel), is so configurable and is such great performer (etc, Debian, etc, etc).
However, I can't refute your above comment due to quirky things I notice from time to time.
Just lately, I've seen a Debian 2.2 based machine I built for a client, get it's nickers in a knot with lots of real and swap memory usage, to the point where I'd like to reboot it after 5pm (having run nicely for 3 or 4 weeks). It just shows a presentation done with javascript to automatically show the clients web site employee bio's via Netscape (which seems to be the root cause of the memory munching). But instead of being able to reboot the machine, it often comes to what seems to be a complete halt due to low memory which is not completely exhausted.
Try cat/dev/urandom and watch your terminal get messed up with wrong chars after you Ctrl-C the cat. FreeBSD does'nt do that.
While trying to figure out WTF is wrong with Linux RAID-0 lately (my transfer rates are *slower* for my striped md devices than their hd devices by about 18%! I used to get a 50% or so increase with 2.2 kernels?!), I tried dd count=1073741824 if=/dev/zero of=/dev/null to see how quick 1Gb of nothing (bottleneck wise, comparitively speaking) would copy to (practically) nowhere, to see if limits I was seeing in my md transfer tests were somehow not related to the md part of the kernel, and sure enough, the resulting transfer I saw was MUCH lower than I was expecting. I thought on my PII-300 I would see a few hundred MB/s, instead? 34MB/s which is not much higher than my single hdd transfer rates. So I try the same thing in FreeBSD... ~250MB/s. Are these not comparable? Is this not a good test or does it not really show the highest I could see from a disk since *this* could be *a* bottleneck?
Can these things be explained and perhaps fixed if need be? I used FreeBSD 4.4 Release, and am planning on installing it proper with a view to perhaps moving my primary desktop to FreeBSD. I've also been using OpenBSD since 2.5 and find it extremely polished and professional.
I love Linux stability, flexibility and all, but against something like a BSD or Solaris, it seems a little flakey and unprofessional. Someone please prove me wrong.
The BSD licence is different from the GPL enough that moral flame wars ensue from time to time. I'm glad we have the freedom as far as beer and speech goes, but I'm not a passionate advocate of one over the other. The GPL seems to bias 'freedom' towards end users (assuring code gets better for all) whereas the BSD licence seems to bias 'freedom' towards coders (protecting a coders IP if they wish it to be kept in the dark). As far I as know it, both allow both types of freedom (beer and speech), but the GPL demands that any GPL source code used or extended which is to be provided in binary form, must also provide the source to that program, extensions and all. Whereas, the BSD licence seems to allow the distribution of binaries without source code (extended or not), as long as credit is given for the authors of the original code used.
I'm probably wrong about something here and will now cop an earfull!...
Yeah, you would'nt want to use an OS maitained by someone who actually enforces their policies in the name of system security. Someone who won't allow their OS to be softened by a few people who all of a sudden don't share ideals that have been intergral to the OS since dot. If they don't like Theo's rationale, then tough.
An OS of the calibre of OpenBSD cannot exist without a leader with sound fundamental ideals who won't allow them to be changed by another ego. If Theo has a big ego, good, he can be forgiven given the quality of OpenBSD.
I hear XP is out now, maybe that is an avenue for you.
In '96 I was working for a stock exchange. We had, seperate networks for production, development and office workers.
Prod and Dev had machines ranging from big DEC VAX'es running DEC VMS and DEC Unix, to multi CPU DEC Alpha's running same plus WinNT.
So, that level of care is not common to MS shops and uncommon to us Unix zealots. What was definite though, was that the servers running WinNT were nowhere near as reliable as the Unix and VMS servers. Including those on the same hardware (Alphas).
BTW, we had a Y2K team, working on fixing date issues that could be demonstrated to bring our trading systems to its knees, back in 1996. Which angers me when people say Y2K was overhyped or a myth (It was'nt overhyped, it was mostly fixed).
What makes you think the US won't use those experiences?
This won't be the same sort of war.
The US will most likely start with a massive bombing campaign, blowing up military, government (and "accidentaly" civilian) support structures with laser guided GBU-15's, etc. Electricity stations, telephone exchanges, fuel depots, water supplies, hospitals ("whoops"), barracks, destroying runways, ammo stores, etc. They just won't stop until all the enemy soldiers are hungry, weak and ill-equiped.
That terrain is perfect for AH64 Apache's shooting hellfire over mountains with a ground troop painting targets with his laser mounted rifle.
They may be good with their AK's (which probably don't have much rifling left, what with all the joyous shooting full-auto into the air every five minutes), but they won't know what hit them when ally soldiers with night vision are shooting at night. Allied soldiers with digital crypto frequency hopping comms, satalite links, super accurate GPS, spy satelite info, etc.
How could the Soviets win with low tech and low numbers. This is not the same type of war, and soon, Afganistan, will be known as Pakistan.
Ask a question of extreme technical and political complexity, to people who have no clue of the in's and out's of the technology or the corruption of their.gov, and then the laws will pass and they won't know what the fuck hit them.
The land of the free, is soon to be no more (not that it ever really was).
I wonder how feasible internet connected CD burning vending machines could be? Perhaps strategically placed at Universities, large train stations and malls.
Remotely administrated, for downloading of latest ISO's of the BSD's, Linux, QNX, BeOS, OSS softwares, etc. Accepting credit card orders on the net, allowing the user to choose where they would like to pick the CD up if they wish it to be pre-burnt and held, or perhaps even keeping a pre-burnt minimum of the latest, most popular for instant purchases. Or perhaps 1900 number purchasing with a mobile phone through a telephone voice menu system that asks for vending machine number, and requested CD(s), the cost of the call paying for the CD and the automated call centre authorising the machine to dispense.
Just to cover the cost of admin, machine maintenance, media, electricity and net connection, etc. Would people spend 1 or 2 bucks for the convenience of 20 seconds in front of a vending machine on their way home?
The machine could alert admin when the cdr low water mark is reached, hardware faults, etc. Keep stats on most popular images etc.
I've been toying with this idea for a while, just wish I had the money to try it out. Perhaps this could also be the future of book purchases (then again, by that time, everyone will probably have broadband? Assuming we live through WW3).
but one thing that open source people haven't learned that MS and Apple and such have learned are to answer the following questions:
People who are most likely to use FreeBSD, are more likely to test for themselves, rather than just blindly believe what some press release tells them or just keep using what works for them and only patch security and stability problems as they arise.
1. is it faster?
Typical question from an MS user (can you blame them?), but nix users tend to be practical people, interested more in flexibility and stability. If it works for them, they keep using it.
2. does it do more/kewler stuff?
cat/usr/share/dict/words|grep -ixc kewl
0
Maybe I need an MS spell checker?
3. will it crash less frequently?
Less frequently than practically never?
4. will it boot faster?
You think most people here care about how fast their OS boots?
5. will i still have to spend hours trying to install new programs and hardware?
You? Probably.
6. does it come with new/more/kewl goodies like MS Office (or equivalent),
Yeah, because computer science is just not science without Microsoft Office, eh?
a dictionary and thesaurus,
I quite like gdict thanks.
100 free hours of internet access, etc.?
Yeah, because what is an OS without a hook into some crap ISP that demands credit card details for a free service that you'll need a top lawyer to get out of after the free bit ends. Chances are, those ISP's are built with a free nix like FreeBSD.
only when an open source OS states these things in their press release will the general public listen.
Whatever. Talk to the hand, 'cause the face aint list'nin.
I'm sure 3M was making a mouse exactly like this one that was on the.au market years ago.
Try moving your mouse around by resting your semi closed hand on it. Notice how much more effort that is required and what an uncomfortable method this would be. I don't move my mouse with my arm, I flick it around the entire screen, and have done so for more than 10 years starting with the original MS mouse (the pregnant ergo MS mouse is horrible for this though, the bump gets in the way), with three fingers moving with my wrist stationary on the mouse pad.
This 3M joy-mouse is not even optical! I rather love my Logitech Optical Mouseman Wheel, I don't even wish it were a wireless one, as I have the mouse cable, cable-tied with enough slack in a loop, to my keyboard cable where it enters the keyboard case, this way, it never gets caught or drags on anything (on a keyboard/mouse slide drawer). It may as well be wireless, since the days of feeling the cable rub and catch on things are gone.
I only wish it had a much higher sampling rate, so as to avoid what seems to be phasing effects that cause the opposite movement of what is done, when moved quickly and suddenly. It anoys me in Starcraft games sometimes!
Just two examples, from one issue of a PC magazine, that makes my blood boil.
I bought this mag, because it had a review of the notebook that I was seriously considering purchasing at the time (Dell Inspiron 8000).
I'll leave the best till last and start with Pg.79, "Platypus QikDRIVE2":
David Lin, is reviewing a PCI card that can hold up to 2Gb of standard PC100 ECC SDRAM in a battery backed setup that interfaces to the PC through a built in SCSI controller as a super quick hard drive (which does not support booting, I guess due to a limited SCSI design. No BIOS?).
Now I agree, that this product could be neato in some situations, like having OS' load really fast, if it were not so brain dead that it cannot boot.
Bear in mind, that as reviewed, this product is $6,000 for 1Gb of battery backed RAM accessable as a SCSI disk. At the moment, in Sydney I can purchase 1Gb of non ECC PC133 SDRAM for $288 inc tax.
He sees the value of this product like this...
"VERDICT Expensive but satisfying new lease of life on I/O bottlenecks, particularly for high volume transaction servers."
It even supports "Free BSDi".
"Where the QikDRIVE is designed to be most effective is in application specific-instances where rapid I/O is required."
He then makes out that this is great for a Win9x machine so you can place your swap file on it! To "scientifically" prove the worth of this product, he compares a Windows 2000 machine with only 128Mb RAM, with and without the severely hampered 1Gb SDRAM, finds that it delivers a 266 times improvement for 2kb transactions, 40 times for 64k streaming write transactions and over 100 times improvement for web server transactions.
I wish I could have been there to pull that RAM off the card and put it into the bloody motherboard to see his results and face then! When the swap file should hardly be required, and his scientific results lurch upwards even more now that this RAM is no longer slowed by about 10 times by the PCI bus and more still by SCSI itself, which is designed to improve inherently slow disks rather than impact the performance of extremely fast (by comparison) RAM.
He then continues with a file copy test, that yields seemingly amazing results (to him at least)! ; ) He copies 120Mb worth of compressed files from the hard disk (ATA/66) to the QikDRIVE in 5 seconds, yielding a 24Mb/s transfer. Wow he thinks with this, "While these sorts of benchmark readings can be pie in the sky lies, just a simple copy and paste exercise demonstrates the speed of the QikDRIVE2".
Ahhhh, Dave, you just witnessed the speed of your hard drive, not the speed of the QikDRIVE. And whats more, did you previously compare a "cut and paste" from and to the same hard drive to compare? An IDE drive at that!?!?!
If he tried a raw read from the QikDRIVE, he may have seen around 80Mb/s or so. Hell, I can copy 120Mb from one drive to another in Linux (not comparing OS', merely what I use) in 5 seconds also!
It amuzes me that it says in the verdict, that this is a "new lease of life on I/O bottlenecks", whilst describing 1Gb of SDRAM, restrained by the PCI bottleneck and slowed by SCSI!
At the end, in typical UK mag style, they give it some completely meaningless ratings. In this case 6 exclamation marks out of 6 for performance! 5/6 for features, 5/6 for value for money and 5/6 overall.
HUH!? 5/6 for value for money, for 1Gb SDRAM costing SIX THOUSAND DOLLARS!? How much was the bloody battery!?
"What a difference a bundle of RAM on a PCI card can make to a server!" ; )
And now here's something I hope you'll really like, Pg.180, "Duplex conundrum revisited Just how much bandwidth should you be getting through your network?":
Now this guy is very aptly described (by the editor?) near a little B&W photo of himself as, "Steve Cassidy Can be found pretending he knows what he's doing with networks, from San Tropez to a factory in Dallas. He can be reached at scassidy@pcauthority.com.au"
Our mate Steve seems to also have trouble understanding the ins and outs of bottlenecks and throughput, but that does'nt stop him from belittling his users who are "curiously unqualified to comment upon"! Hey? Because they are just "users", they are not qualified to speak about this dickheads badly performing network? Even though they can most likely speak comparatively.
I can't be bothered to quote too much of this guys crap, but he's having some troubles with 100Mbit full-duplex with IPX/Netware. He watches how the machines set to half-duplex perform to his trained eye, faster than the F-D PC's whilst going through a Netware login script. "This is the type of skill that network managers develop without even knowing that it sets them apart from the users they're trying to support", God I'm going to vomit soon.
"it's a bit like knowing enough about modems that you can hear a bad X2 or V.90 entrain sequence, even while the machine is trying to connect." Thats retrain Steve. My mother knows when her modem does'nt complete it's connect sequence!
"When `the network goes slow' it requires considerable people-management skill to explain to them whats going on", ahh better yet, why not use some technical skill to fix the network?
"When someone asked me recently just what throughput he should expect from a 100Mbits/sec LAN, I was struck somewhat dumb. Should I tell him what I was seeing here, or at those clients who spent their hard-earned money on HP managed hubs? Might I be revealing rather too much if I came up with a throughput number that was lower than his?" Why not give him the math and then explain ethernet and ipx protocol overhead coming into play on your shared bandwidth hubs.
"In the end, I decided to publish and be damned, and presented him the numbers I see from my regular tape backup jobs, using a wide range of kit in all manner of odd network configurations". Sounds like weird science to me!
"A basic Pentium/166 sitting inside a very venerable Compaq ProLiant backing up to DLT manages just over 90Mb per minute, working locally - any limitation caused by the slow CPU is offset by the fact that the communication is local and it's still at the lower end of the range of tape backup speeds." Slow CPU? Huh? This guys rekons that a Pentium/166 is starts to show it's limits at around 12Mbits/sec? This guy is an "authority" in computer science networking, bottlenecks and throughput? Hell Steve, the PCI bus is 1064Mbit/sec! I mean fuck, my old 340Mb Western Digital Caviar EIDE drive can transfer data at that rate with my i486DX-33, not that the CPU is relevant in this case or yours!
"Most modern kit can manage to hit the 95Mb to 110Mb per minute bracket when backing up over a gigabyte of stuff", the tape drives perhaps, but thats not a measure of 100Mbit capabilities.
"So, the short answer is that a 100Mb file moving to or from a server on a 100Mbits/sec LAN should move in around a minute. Any slower and you should suspect that something's hogging your bandwidth, or disracting the processors at either end, or your ethernet cards have a personality clash with your hub" Ahh haaa, oh-kay. 100Mbit=13.3Mbit, yeah, makes complete sense. On my 100Mbit network, 100Mbytes copies, across a dedicated connection around a bit under 10 seconds.
"Don't worry about how the maths can lead from what should be 6 billion bits in 60 seconds down to one hundred million eight-bit bytes moving in the same period - this is the real world and those are real-world numbers, recorded over about three months of nights in backup log files (I must confess that I have these systems set to print out at job-end, and junior staff encouraged to file those print-outs in antedilivian ring-binders.) Lots of you evidently want to be able to attach practical numbers to the very smoke-and-mirrors estimates from the industry, so there's a practical figure to start thinking about." Can you beleive this moron? I think he actually thinks he is a networking guru!
I love this one... "In a spirit of perversity, I installed the Netgear dual-Gigabit Ethernet 618 switch here and dropped the matching copper gigabit cards into my ancient dual-Pentium Pro/200 ALR server. On paper, even though the ALR has LVD disks, there should have been no headroom available from those poor old CPU's to perk up LAN performance in the gigabit environment - how could a system bus running at 66MHz cope with data travelling at one billion bits/sec?" Take your MHz and multiply by bus width Steve. A 33MHz PCI bus runs that quick!
But this takes the cake!... "But it is quicker: trust me, I'm a nerd."
Morons like this, should NOT be writing about technical aspects of anything! If someone was learning about this stuff and did'nt know any better, they might beleive the crap that comes from this guy, who rekons he's a bit of a propeller head.
A typical moron working in IT who's trying to fool everyone that he knows his shit and everyone else around him does not. The scarey thing is, that this arsehole probably has all the ridiculous certification that the idiotic industry holds in such high regard.
All you have to do is fdisk your drive, mark hda1 as Linux swap (0x82),
Why do I have to mark the only blank partition (intended for Solaris) as a Linux swap partition? I already have one on each of hda and hdc (both priority 1).
If the Solaris 8 install was half decent, it would not spit the dummy just because it can't see the whole drive as potentially its own. There is free space there, that all the free OS' can install to.
and boot from Software CD #1 instead of the Installation CD.
How silly of me to expect the install from the install CD to actually work with on an IDE drive that is not only the primary master, but also the 1st partition.
If you had even made a cursory search for this on USENET under alt.solaris.x86, you would have easily found the answer.
Thanks for pointing it out though. Not that it makes the installer any better though.
Slashdot Mirror
Score:5, Informative? (No offence Michael) People who moderate Informative are by default ignorant, or else it would'nt be informative to them. So of course, they are not qualified to say this is "Informative" without your post having supporting evidence, links, etc.
The thought that it's "not the drives but bad handling" is silly. There are bad and good batches. It seems from this story that there might be a bad batch of IBM's. Do you think that mostly these IBM hdd owners for some unknown reason are handling or treating their drives baddly whereas the other owners are more enlightened on drive care? Not a chance.
Poor PSU quality, heat removal, shock, vibration and simply bad manufacturing batches or even poor design (rare to get past Q-Ctrl before mass mfg) can be the causes of drive failure. The problem CAN be the drives and in this case seems to be.
http://www.driveservice.com/bestwrst.htm
The damage done by a bump is usually not evident for months until the drive fails. Apparently a drive will get a small mechanical irregularity, and then slowly chew on itself until failure.
An impact strong enough to cause the hovering heads to actually touch the moving platters can cause the paint like magnetic surface to fleck, causing bad blocks. In time, as the heads pass over that area, they can cause further chipping off of the surface and thus the bad blocks grow in numbers. This may result in data loss and possibly destruction of heads alltogether.
With modern drives, that automatically re-map defective blocks to spares set aside to guarantee maximum usable blocks, this problem can be masked until those spare blocks are exhausted.
You need a high enough impact or vibration to cause a head to come down onto a spinning platter.
If you put a tower case on the floor, and the floor moves a little every time someone walks near, expect problems.
Vibration levels of a person walking near a PC on the floor is nowhere near high enough to cause any problem. The vibration induced by the drives own head movement will cause far quicker acceleration and total drive movement than that (which is mostly a safer horizontal movement). Unless the PC is on creaky floorboards, but even then this is *slow* movement.
If you put a case on a concrete floor, but it is often knocked during the day, expect failure.
Yeah, if the PC itself is knocked, it could cause drive problems, depending on what knocked it (how hard, heavy and fast the object was moving). But I imagine people who care for thier PC's actually take care of them.
If a computer is on a table that moves a little while you are working, it may not last long.
The table is bound to take a lot of the usual day to day shock, which should be quite minimal. Sure if you're dropping heavy objects onto your desk with your PC on it then you might be at risk. But not typing, mousing, paper folders, general office duties type stuff or even dropping a book onto your table. The table stops the book, not the hdd, and any vibration that gets to the hdd would be very damped by the desk.
This failure mode is dependent on how much movement about the axis actually happens, of course.
Which axis? Drives don't like to be yawed while they are spinning as the gyroscopic effect puts strains on the bearings in ways they were'nt designed for. Sideways shock should cause minimal problems, it is the vertical shock that causes heads to crash.
Drives are built to handle a lot of Gs when they are not powered, but when they are running they are very vulnerable.
Very true. Usually 4 or 5 times. Drives are also more susceptible to vibration than shock. My drives takes 300Gs shock non-operating, 63 operating, but only 0.5Gs vibration while operating!
Put a drive on its own power supply connector.
Every PSU I have ever had cause to open, had all it's GND, +5V and +12V wired to common power rails within the PSU. Seperate connectors barely make a difference. Anyway, drives tend to not use much power once they have reached their stated spindle speeds after power up. My drive draws 27 Watts during power-on, then 11.6 during seeks, 8.5 reading/writing and only 7 Watts during idle (platter spinning, no head movement or RW). This is a 7200 rpm drive and I have two of them.
I guess one possible saving grace at boot time is that the CPU is not doing a real lot while the hdds are spinning up. As an interesting comparison, my Intel Pentium II 300MHz, draws 43 Watts at peak!
I've had good luck with considerable quantities of Western Digital drives. Good support, also. I've had bad luck with Quantum, Seagate, and Maxtor.
Problem is, that ALL manufacturers have particularly bad batches and good batches. I've personally had good experience with WD, IBM and Seagate, so-so Quantum and absolutely dismal experience with Maxtor (I will never buy one again). WD is known to have past models that do NOT adhere to standards that the drive reports to support.
Drive manufacturers in the past (not sure if they still do), sold "special" models that passed as a grade A drive, built to higher tolerances (almost as if mil-spec) and sold for higher prices for server usage. A DEC rep told me this years ago and looking at the lifetimes and warrantee periods of the high end SCSI drives we were using, I would tend to beleive him.
Recently, while trying to find out why Linux 2.4.x software RAID-0 device performance was about 18% SLOWER than the /dev/hd# devices that make up the stripe, I decided to do a simple little test from single user mode that I thought might show the maximum read speed I could expect from my Linux kernel (2.4.10 at the moment, as tested).
/dev/hda and /dev/hdc partitions to set up FreeBSD software RAID-0 to see if I should move my desktop #1 from Linux to FreeBSD.
So I did a (probably very unscientific) # dd if=/dev/zero of=/dev/null with a count= of 1Gb, timed it and worked out the transfer rate... ~34M bytes/sec. WTF?!?!
I then did the same in FreeBSD 4.4 Release.................... 250M bytes/sec!
I'm now off to wipe some adjacent
PS, I have a PII-300 (100MHz FSB) with 256Mb PC100 SDRAM for comparison.
Troll...Do you live in...how about telling the story as it really is...just remember the damn thig was there 20 years before anyone esle...be in Sydney...70 k away...You or your parents...
What a great big steaming, anonymous, assuming pile of horse shit.
I have lived in Bondi for almost 30 years. My mother and her father also at a nearby beach for much longer than that.
Lucas Heights is about 28km from the Sydney CBD.
I prefer that a "fairly safe reactor" be way out past Dubbo than 28km from SYDNEY CBD. Especially when the new proposed site docco states "the cut-off dose is reached 25 km from the proposed site", they also speak of doses out to 50km in an accident situation being well under safe levels. Hey I've got an idea, why don't they just put it 500km inland!?
Are the residents of the Leichhardt, Ashfield, Strathfield, Fairfield, Bankstown, Marrickville, Canterbury, Rockdale, Hurstville and Kogarah areas to be blamed for living within the 25km unsafe zone also?
"Fairly safe" is not safe enough.
Oh man, it is rare for me to enjoy a hollywood flick. They usually have some bullshit device, ability or random buzzwords to help patch up an awful script. Which is especially annoying when you recognise the buzzwords or see that parts of the movie are extremely over explained with that shit for the stupid audience.
I know they're often meant to be amazing to watch but fuck. Swordfish, Broken Arrow, Face Off, Anaconda, MI2, etc etc etc.
I enjoyed Alien, The Matrix, The Sixth Sense, Shrek and a few others I can't remember right now, but most of what I see I think I could get more entertainment out of masturbating with a handfull of rusty razor blades.
I like sci-fi, but I don't when I am treated like an idiot with over simplified bullshit. If they have to describe something non-existant, at least make up a new word for it! Instead of "ip-chains multiplexor" for a time machine, or some shit like that.
Anyone know of any web sites that quote the bullshit from movies?
Plus, imagine if those planes were to be ploughed into reactors around the US.
I live in Sydney, we have this tiny little reactor that is for medical use. Regardless I'm not real pleased with the fact that IT WAS a target during the games and probably will be forever more, until a sane government decides to shut it down and relocate to a safer area away from innocents.
I suppose you'll get used to that lightness after while but it just feels odd.
I used to like the original MS mouse for it's weight, but now I have an optical Logitech wheel mouse, when ever I use someone elses mouse I find it feels too heavy and usually jumpy.
Although the high speed tracking problem can cause problems at times, I find nothing as smooth (both physical feel and tracking) and love the light weight.
I have a so called 2nd generation optical though.
A mate and I set up a Quake2 party complete with a digital projector set up on a seperate machine on the network using the "camera where the action is" mode and..... no bloody one came. Everyone wanted to go see anyother stupid hollywood flick. :(
I really, really, really doubt this.
I remember reading that the GSM StarTAC showed within it's secret network menu, Mobile-Base distance in metres. However, in the interest of correct info and to the detriment of my karma, I delved into the GSM specs and found...
That the value that can be used to calculate distance is a 6 bit number that divides up the max distance within the 35.2km usable time window, giving an accuracy from one sample (one cell) of about 550 metres.
So your value of 400m for multiple cells is probably correct. One thousand appologies.
In Australia, with GSM, you call 000 or the international GSM emergency number (I forgot it), and it will automatically call the Fire/Ambo/Police number through your network provider (Telstra, Optus, Vodaphone or a reseller), if you have no signal to your own provider it will allow usage of any provider you have signal with for your emergency call. Even without a SIM card inserted in the phone. This is a legal requirement and seems to be a feature built in to GSM itself.
Due to the very precise time division multiplexing used with GSM, the distance you are from the base station you are currently subscribed can be gleaned down to a metre. If they can force your phone to switch to 2 other cells after an emergency call, they could probably pin point you without GPS. With the hidden Network menus in Motorolla StarTac GSM and Nokia phones, you can see how far you are from the base station in metres.
there are other reasons why many still consider Linux a toy.
/dev/urandom and watch your terminal get messed up with wrong chars after you Ctrl-C the cat. FreeBSD does'nt do that.
I've been using Linux for about 4 years, and advocate it based on the fact that it rarely (once or twice) crashes on me (the kernel), is so configurable and is such great performer (etc, Debian, etc, etc).
However, I can't refute your above comment due to quirky things I notice from time to time.
Just lately, I've seen a Debian 2.2 based machine I built for a client, get it's nickers in a knot with lots of real and swap memory usage, to the point where I'd like to reboot it after 5pm (having run nicely for 3 or 4 weeks). It just shows a presentation done with javascript to automatically show the clients web site employee bio's via Netscape (which seems to be the root cause of the memory munching). But instead of being able to reboot the machine, it often comes to what seems to be a complete halt due to low memory which is not completely exhausted.
Try cat
While trying to figure out WTF is wrong with Linux RAID-0 lately (my transfer rates are *slower* for my striped md devices than their hd devices by about 18%! I used to get a 50% or so increase with 2.2 kernels?!), I tried dd count=1073741824 if=/dev/zero of=/dev/null to see how quick 1Gb of nothing (bottleneck wise, comparitively speaking) would copy to (practically) nowhere, to see if limits I was seeing in my md transfer tests were somehow not related to the md part of the kernel, and sure enough, the resulting transfer I saw was MUCH lower than I was expecting. I thought on my PII-300 I would see a few hundred MB/s, instead? 34MB/s which is not much higher than my single hdd transfer rates. So I try the same thing in FreeBSD... ~250MB/s. Are these not comparable? Is this not a good test or does it not really show the highest I could see from a disk since *this* could be *a* bottleneck?
Can these things be explained and perhaps fixed if need be? I used FreeBSD 4.4 Release, and am planning on installing it proper with a view to perhaps moving my primary desktop to FreeBSD. I've also been using OpenBSD since 2.5 and find it extremely polished and professional.
I love Linux stability, flexibility and all, but against something like a BSD or Solaris, it seems a little flakey and unprofessional. Someone please prove me wrong.
The BSD licence is different from the GPL enough that moral flame wars ensue from time to time. I'm glad we have the freedom as far as beer and speech goes, but I'm not a passionate advocate of one over the other. The GPL seems to bias 'freedom' towards end users (assuring code gets better for all) whereas the BSD licence seems to bias 'freedom' towards coders (protecting a coders IP if they wish it to be kept in the dark). As far I as know it, both allow both types of freedom (beer and speech), but the GPL demands that any GPL source code used or extended which is to be provided in binary form, must also provide the source to that program, extensions and all. Whereas, the BSD licence seems to allow the distribution of binaries without source code (extended or not), as long as credit is given for the authors of the original code used.
I'm probably wrong about something here and will now cop an earfull!...
Yeah, you would'nt want to use an OS maitained by someone who actually enforces their policies in the name of system security. Someone who won't allow their OS to be softened by a few people who all of a sudden don't share ideals that have been intergral to the OS since dot. If they don't like Theo's rationale, then tough.
An OS of the calibre of OpenBSD cannot exist without a leader with sound fundamental ideals who won't allow them to be changed by another ego. If Theo has a big ego, good, he can be forgiven given the quality of OpenBSD.
I hear XP is out now, maybe that is an avenue for you.
in addition to patch-rml-2.4.10-preempt-kernel-1
Grrr, thanks. Late night last night. ; )
Basically, why is this not the default behavior?
My clean 2.4.10 kernel patched with this does not compile, giving me undefined reference in kernel.o, do_signal and syscall_trace functions.
Maybe I missed some config setting?
Disclaimer: I work in a Windows shop
Not that Windows shops came up with those ideas.
In '96 I was working for a stock exchange. We had, seperate networks for production, development and office workers.
Prod and Dev had machines ranging from big DEC VAX'es running DEC VMS and DEC Unix, to multi CPU DEC Alpha's running same plus WinNT.
So, that level of care is not common to MS shops and uncommon to us Unix zealots. What was definite though, was that the servers running WinNT were nowhere near as reliable as the Unix and VMS servers. Including those on the same hardware (Alphas).
BTW, we had a Y2K team, working on fixing date issues that could be demonstrated to bring our trading systems to its knees, back in 1996. Which angers me when people say Y2K was overhyped or a myth (It was'nt overhyped, it was mostly fixed).
What makes you think the US won't use those experiences?
This won't be the same sort of war.
The US will most likely start with a massive bombing campaign, blowing up military, government (and "accidentaly" civilian) support structures with laser guided GBU-15's, etc. Electricity stations, telephone exchanges, fuel depots, water supplies, hospitals ("whoops"), barracks, destroying runways, ammo stores, etc. They just won't stop until all the enemy soldiers are hungry, weak and ill-equiped.
That terrain is perfect for AH64 Apache's shooting hellfire over mountains with a ground troop painting targets with his laser mounted rifle.
They may be good with their AK's (which probably don't have much rifling left, what with all the joyous shooting full-auto into the air every five minutes), but they won't know what hit them when ally soldiers with night vision are shooting at night. Allied soldiers with digital crypto frequency hopping comms, satalite links, super accurate GPS, spy satelite info, etc.
How could the Soviets win with low tech and low numbers. This is not the same type of war, and soon, Afganistan, will be known as Pakistan.
Sorry about #5 kootch.
I'm in a bad mood and the thought that FreeBSD needs MS Office set me off.
: )
Ask a question of extreme technical and political complexity, to people who have no clue of the in's and out's of the technology or the corruption of their .gov, and then the laws will pass and they won't know what the fuck hit them.
The land of the free, is soon to be no more (not that it ever really was).
Look at the bunny! Look, see the cute bunny!?!?
I wonder how feasible internet connected CD burning vending machines could be? Perhaps strategically placed at Universities, large train stations and malls.
Remotely administrated, for downloading of latest ISO's of the BSD's, Linux, QNX, BeOS, OSS softwares, etc. Accepting credit card orders on the net, allowing the user to choose where they would like to pick the CD up if they wish it to be pre-burnt and held, or perhaps even keeping a pre-burnt minimum of the latest, most popular for instant purchases. Or perhaps 1900 number purchasing with a mobile phone through a telephone voice menu system that asks for vending machine number, and requested CD(s), the cost of the call paying for the CD and the automated call centre authorising the machine to dispense.
Just to cover the cost of admin, machine maintenance, media, electricity and net connection, etc. Would people spend 1 or 2 bucks for the convenience of 20 seconds in front of a vending machine on their way home?
The machine could alert admin when the cdr low water mark is reached, hardware faults, etc. Keep stats on most popular images etc.
I've been toying with this idea for a while, just wish I had the money to try it out. Perhaps this could also be the future of book purchases (then again, by that time, everyone will probably have broadband? Assuming we live through WW3).
but one thing that open source people haven't learned that MS and Apple and such have learned are to answer the following questions:
/usr/share/dict/words|grep -ixc kewl
People who are most likely to use FreeBSD, are more likely to test for themselves, rather than just blindly believe what some press release tells them or just keep using what works for them and only patch security and stability problems as they arise.
1. is it faster?
Typical question from an MS user (can you blame them?), but nix users tend to be practical people, interested more in flexibility and stability. If it works for them, they keep using it.
2. does it do more/kewler stuff?
cat
0
Maybe I need an MS spell checker?
3. will it crash less frequently?
Less frequently than practically never?
4. will it boot faster?
You think most people here care about how fast their OS boots?
5. will i still have to spend hours trying to install new programs and hardware?
You? Probably.
6. does it come with new/more/kewl goodies like MS Office (or equivalent),
Yeah, because computer science is just not science without Microsoft Office, eh?
a dictionary and thesaurus,
I quite like gdict thanks.
100 free hours of internet access, etc.?
Yeah, because what is an OS without a hook into some crap ISP that demands credit card details for a free service that you'll need a top lawyer to get out of after the free bit ends. Chances are, those ISP's are built with a free nix like FreeBSD.
only when an open source OS states these things in their press release will the general public listen.
Whatever. Talk to the hand, 'cause the face aint list'nin.
I'm sure 3M was making a mouse exactly like this one that was on the .au market years ago.
Try moving your mouse around by resting your semi closed hand on it. Notice how much more effort that is required and what an uncomfortable method this would be. I don't move my mouse with my arm, I flick it around the entire screen, and have done so for more than 10 years starting with the original MS mouse (the pregnant ergo MS mouse is horrible for this though, the bump gets in the way), with three fingers moving with my wrist stationary on the mouse pad.
This 3M joy-mouse is not even optical! I rather love my Logitech Optical Mouseman Wheel, I don't even wish it were a wireless one, as I have the mouse cable, cable-tied with enough slack in a loop, to my keyboard cable where it enters the keyboard case, this way, it never gets caught or drags on anything (on a keyboard/mouse slide drawer). It may as well be wireless, since the days of feeling the cable rub and catch on things are gone.
I only wish it had a much higher sampling rate, so as to avoid what seems to be phasing effects that cause the opposite movement of what is done, when moved quickly and suddenly. It anoys me in Starcraft games sometimes!
will probably kick it's citizens arses for so much as rot13'ing their emails.
Just two examples, from one issue of a PC magazine, that makes my blood boil.
I bought this mag, because it had a review of the notebook that I was seriously considering purchasing at the time (Dell Inspiron 8000).
I'll leave the best till last and start with Pg.79, "Platypus QikDRIVE2":
David Lin, is reviewing a PCI card that can hold up to 2Gb of standard PC100 ECC SDRAM in a battery backed setup that interfaces to the PC through a built in SCSI controller as a super quick hard drive (which does not support booting, I guess due to a limited SCSI design. No BIOS?).
Now I agree, that this product could be neato in some situations, like having OS' load really fast, if it were not so brain dead that it cannot boot.
Bear in mind, that as reviewed, this product is $6,000 for 1Gb of battery backed RAM accessable as a SCSI disk. At the moment, in Sydney I can purchase 1Gb of non ECC PC133 SDRAM for $288 inc tax.
He sees the value of this product like this...
"VERDICT Expensive but satisfying new lease of life on I/O bottlenecks, particularly for high volume transaction servers."
It even supports "Free BSDi".
"Where the QikDRIVE is designed to be most effective is in application specific-instances where rapid I/O is required."
He then makes out that this is great for a Win9x machine so you can place your swap file on it! To "scientifically" prove the worth of this product, he compares a Windows 2000 machine with only 128Mb RAM, with and without the severely hampered 1Gb SDRAM, finds that it delivers a 266 times improvement for 2kb transactions, 40 times for 64k streaming write transactions and over 100 times improvement for web server transactions.
I wish I could have been there to pull that RAM off the card and put it into the bloody motherboard to see his results and face then! When the swap file should hardly be required, and his scientific results lurch upwards even more now that this RAM is no longer slowed by about 10 times by the PCI bus and more still by SCSI itself, which is designed to improve inherently slow disks rather than impact the performance of extremely fast (by comparison) RAM.
He then continues with a file copy test, that yields seemingly amazing results (to him at least)! ; ) He copies 120Mb worth of compressed files from the hard disk (ATA/66) to the QikDRIVE in 5 seconds, yielding a 24Mb/s transfer. Wow he thinks with this, "While these sorts of benchmark readings can be pie in the sky lies, just a simple copy and paste exercise demonstrates the speed of the QikDRIVE2".
Ahhhh, Dave, you just witnessed the speed of your hard drive, not the speed of the QikDRIVE. And whats more, did you previously compare a "cut and paste" from and to the same hard drive to compare? An IDE drive at that!?!?!
If he tried a raw read from the QikDRIVE, he may have seen around 80Mb/s or so. Hell, I can copy 120Mb from one drive to another in Linux (not comparing OS', merely what I use) in 5 seconds also!
It amuzes me that it says in the verdict, that this is a "new lease of life on I/O bottlenecks", whilst describing 1Gb of SDRAM, restrained by the PCI bottleneck and slowed by SCSI!
At the end, in typical UK mag style, they give it some completely meaningless ratings. In this case 6 exclamation marks out of 6 for performance! 5/6 for features, 5/6 for value for money and 5/6 overall.
HUH!? 5/6 for value for money, for 1Gb SDRAM costing SIX THOUSAND DOLLARS!? How much was the bloody battery!?
"What a difference a bundle of RAM on a PCI card can make to a server!" ; )
And now here's something I hope you'll really like, Pg.180, "Duplex conundrum revisited Just how much bandwidth should you be getting through your network?":
Now this guy is very aptly described (by the editor?) near a little B&W photo of himself as, "Steve Cassidy Can be found pretending he knows what he's doing with networks, from San Tropez to a factory in Dallas. He can be reached at scassidy@pcauthority.com.au"
Our mate Steve seems to also have trouble understanding the ins and outs of bottlenecks and throughput, but that does'nt stop him from belittling his users who are "curiously unqualified to comment upon"! Hey? Because they are just "users", they are not qualified to speak about this dickheads badly performing network? Even though they can most likely speak comparatively.
I can't be bothered to quote too much of this guys crap, but he's having some troubles with 100Mbit full-duplex with IPX/Netware. He watches how the machines set to half-duplex perform to his trained eye, faster than the F-D PC's whilst going through a Netware login script. "This is the type of skill that network managers develop without even knowing that it sets them apart from the users they're trying to support", God I'm going to vomit soon.
"it's a bit like knowing enough about modems that you can hear a bad X2 or V.90 entrain sequence, even while the machine is trying to connect." Thats retrain Steve. My mother knows when her modem does'nt complete it's connect sequence!
"When `the network goes slow' it requires considerable people-management skill to explain to them whats going on", ahh better yet, why not use some technical skill to fix the network?
"When someone asked me recently just what throughput he should expect from a 100Mbits/sec LAN, I was struck somewhat dumb. Should I tell him what I was seeing here, or at those clients who spent their hard-earned money on HP managed hubs? Might I be revealing rather too much if I came up with a throughput number that was lower than his?" Why not give him the math and then explain ethernet and ipx protocol overhead coming into play on your shared bandwidth hubs.
"In the end, I decided to publish and be damned, and presented him the numbers I see from my regular tape backup jobs, using a wide range of kit in all manner of odd network configurations". Sounds like weird science to me!
"A basic Pentium/166 sitting inside a very venerable Compaq ProLiant backing up to DLT manages just over 90Mb per minute, working locally - any limitation caused by the slow CPU is offset by the fact that the communication is local and it's still at the lower end of the range of tape backup speeds." Slow CPU? Huh? This guys rekons that a Pentium/166 is starts to show it's limits at around 12Mbits/sec? This guy is an "authority" in computer science networking, bottlenecks and throughput? Hell Steve, the PCI bus is 1064Mbit/sec! I mean fuck, my old 340Mb Western Digital Caviar EIDE drive can transfer data at that rate with my i486DX-33, not that the CPU is relevant in this case or yours!
"Most modern kit can manage to hit the 95Mb to 110Mb per minute bracket when backing up over a gigabyte of stuff", the tape drives perhaps, but thats not a measure of 100Mbit capabilities.
" So, the short answer is that a 100Mb file moving to or from a server on a 100Mbits/sec LAN should move in around a minute. Any slower and you should suspect that something's hogging your bandwidth, or disracting the processors at either end, or your ethernet cards have a personality clash with your hub" Ahh haaa, oh-kay. 100Mbit=13.3Mbit, yeah, makes complete sense. On my 100Mbit network, 100Mbytes copies, across a dedicated connection around a bit under 10 seconds.
"Don't worry about how the maths can lead from what should be 6 billion bits in 60 seconds down to one hundred million eight-bit bytes moving in the same period - this is the real world and those are real-world numbers, recorded over about three months of nights in backup log files (I must confess that I have these systems set to print out at job-end, and junior staff encouraged to file those print-outs in antedilivian ring-binders.) Lots of you evidently want to be able to attach practical numbers to the very smoke-and-mirrors estimates from the industry, so there's a practical figure to start thinking about." Can you beleive this moron? I think he actually thinks he is a networking guru!
I love this one... "In a spirit of perversity, I installed the Netgear dual-Gigabit Ethernet 618 switch here and dropped the matching copper gigabit cards into my ancient dual-Pentium Pro/200 ALR server. On paper, even though the ALR has LVD disks, there should have been no headroom available from those poor old CPU's to perk up LAN performance in the gigabit environment - how could a system bus running at 66MHz cope with data travelling at one billion bits/sec? " Take your MHz and multiply by bus width Steve. A 33MHz PCI bus runs that quick!
But this takes the cake!... "But it is quicker: trust me, I'm a nerd."
Morons like this, should NOT be writing about technical aspects of anything! If someone was learning about this stuff and did'nt know any better, they might beleive the crap that comes from this guy, who rekons he's a bit of a propeller head.
A typical moron working in IT who's trying to fool everyone that he knows his shit and everyone else around him does not. The scarey thing is, that this arsehole probably has all the ridiculous certification that the idiotic industry holds in such high regard.
Morons hiring morons.
All you have to do is fdisk your drive, mark hda1 as Linux swap (0x82),
Why do I have to mark the only blank partition (intended for Solaris) as a Linux swap partition? I already have one on each of hda and hdc (both priority 1).
If the Solaris 8 install was half decent, it would not spit the dummy just because it can't see the whole drive as potentially its own. There is free space there, that all the free OS' can install to.
and boot from Software CD #1 instead of the Installation CD.
How silly of me to expect the install from the install CD to actually work with on an IDE drive that is not only the primary master, but also the 1st partition.
If you had even made a cursory search for this on USENET under alt.solaris.x86, you would have easily found the answer.
Thanks for pointing it out though. Not that it makes the installer any better though.