Not seen any popups with Opera. I see some mention of using Flash to get around it, to which I reply (for your user CSS file, thanks to the author who I seem to have forgotten):
"The default install of XP refuses to boot off of my >137 GB"
Burn a slipstreamed image of XP+SP2.
"I shouldn't have to regularly reinstall everything due to a flakey registry to begin with"
Um, no, you shouldn't. What the hell are you doing, blasting your DIMMs with gamma rays or something?
"ultimatley, that's what's keeping me from going Linux as well. I know that getting the software up and runing on my machine will be just as bad if not worse than reinstalling XP. At least with XP there are drivers to be found"
I'd rather reinstall my FreeBSD workstation than my XP desktop. Driver wise a GENERIC kernel generally does better than XP in my experience, and reinstalling my apps is just a portinstall away. Better than grabbing DOpus, DigiGuide, Vim, PenguiNet, Opera, Firefox, daemon-tools, nVidia drivers, sound card drivers, cygwin, MPC, VLC, NBPro, ffdshow, Samurize, SpeedFan, Locate32, Nero, Azureus, TweakUI, Synergy, etc, all from different places. The difficulty's mostly reduced from actually going and finding those apps to just making a list. *shrug*, YMMV, swings+roundabouts, etc.
"sorry? OpenBSD wants GUI management information? Am I missing something here?"
RAID management involves things like (re)building arrays, adding spares, etc, which requires access to the management interface. Ignore the word "GUI" here; under Linux you access it using a tool called aaccli; under FreeBSD you can use it under Linux emulation (kinda).
Open source drivers are available and have been for years (if not always in a very usable state; Linux aacraid's only recently stopped being marked EXPERIMENTAL), but they don't include management tools; without them you're basically forced to talk to the card using its BIOS at POST, which is kinda suboptimal if you're trying to use hardware RAID to make a system more reliable.
Ironically AAC's support under FreeBSD at least has been superior to Linux's for quite a while (not so much in the past few months, but certainly for the past couple of years before). We originally bought our cards to run under FreeBSD, and had significant problems migrating to Linux where the aacraid driver liked to fall over every few weeks.
I'd rather use software RAID now. Closed source management tools and unreliable software, hardware and firmware are not things I want anywhere fucking near my data storage subsystems.
"you never hear *nix overclockers complain that their OS crashes all the time do you?"
Yes? Although more normally it's things like gcc throwing tonnes of SEGV's and the odd bit of filesystem corruption. Maybe Windows just has more assert()s.
This is a common mistake. You really should parse out each entry and select based on q-values. What happens when the client says application/xhtml+xml;q=0 ("XHTML is unacceptable"), or even text/html;q=1.0, application/xhtml+xml;q=0.6 ("I prefer HTML, but I can handle XHTML to some extent")? Or is it ok to ignore one spec if you kinda follow another better?;)
Oh, and don't forget to switch your <link rel="stylesheet"> bits with <?xml-stylesheet ?> when you're serving as XML.
Personally, I tend to just stick with HTML 4.01. If I'm going to use XHTML, I like to use an XSLT to transform it to HTML 4.01 for naughty little browsers, which also handily enforces well-formedness in a way that allows for degredation to text/html-served broken XHTML instead of a browser-level parse error.
It's a preview for a reason. And what higher standard are you holding Opera to? Firefox has plenty of bugs and misfeatures of its own, and yes, they do tend to go unfixed for a depressingly long time (up to and including never).
Certainly in my experience as a web developer, Opera's tended to fare better than Gecko and KHTML-based browsers with regard to layout problems and incomplete standards support. Opera's hardly perfect, but it gets closer than most.
These are peaks too; you'll only be hitting them with your laptop plugged into the wall and running cpuburn, otherwise they'll be ticking over in low power mode.
Sadly they don't seem to have gone very far with their lossless service; most of their music is still transcoded from ~384kbps (free-format) MP3's. I've bought a few albums from them as FLAC and been very happy to be able to do so, but most of their selection still doesn't meet my requirements for "good quality" (yeah, so I'm a bit anal and probably would never notice the difference, but that's not the only reason I want lossless).
I still vastly prefer Magnatune; if iTunes were like that, I'd probably be bankrupt.
"You cannot accelerate networking very much because the problem is highly serial."
It's serial at line-level (usually), but that doesn't mean you can't process frames/packets in parallel. At the very least you can offload things like checksums, and process packets for different connections in parallel; you can even do simple things like aligning packet payloads on page boundaries so the OS doesn't have to copy them about needlessly (zero-copy is good when you're talking 100's of MB/s, no?). On top of that, I'm pretty sure you can process multiple packets for a single connection in parallel since you can fill different bits of your socket window from multiple packets at once.
"You have all that CPU for free;"
Er, no, you don't; you just said it's been paid for, it's certainly not free. In the case of a heavily loaded server, you don't want half your CPU doing little more than copying memory between the network and your application, doing tedious packet reassembly, calculating checksums and servicing more interrupts than necessary; you want it running your application code to generate that data.
For sending data, there are no special requirements or capabilities that
the sending NIC must have. The data written to the socket, though, must
be at least a page in size and page aligned in order to be mapped into
the kernel. If it does not meet the page size and alignment constraints,
it will be copied into the kernel, as is normally the case with socket
I/O.
The user should be careful not to overwrite buffers that have been writ ten to the socket before the data has been freed by the kernel, and the
copy-on-write mapping cleared. If a buffer is overwritten before it has
been given up by the kernel, the data will be copied, and no savings in
CPU utilization and memory bandwidth utilization will be realized.
It also mentions some issues with regard to zero-copy receive, which requires help from the NIC to ensure received packet payloads are also page-aligned and >= page size. Such support is predictably very rare.
"Much like RAID5 however, there is a space sacrifice for this extra parity layer."
But it's settable; so if you want to be able to recover fully from losing/corrupting 20% of your backup you just set it to 20% of your backup size, and if you only care about a few minor bit errors or so, you can drop it to a couple of percent or less.
Be nice if vendors provided PAR2's for their ISO/DVD images/anything else big; it sucks when you find the MD5 of your download doesn't match the one they provide (or that 400MB setup.exe throws a checksum mismatch and refuses to run), and you know it's probably just a single bit flipped somewhere but can't do much beyond redownloading the entire thing. rsync helps, of course, but that's a *lot* heavier on the server both in resource use and administration cost.
Funny; the two Maxtor DiamondMax 9's my friend just bought simply don't work properly with the Intel SATA chipset he wanted to use them with. They're fine on the Promise/Via controllers on his desktop, but on the Intel and libata they throw IO errors after about 50 sectors. The two Seagates he bought with them are fine...
A bit of research suggests these drives are prone to firmware bugs and outright failure, but of course Maxtor's site doesn't appear to mention any updates for them.. bah.
The SATA ones are a little buzzier than the PATA ones (they have different acoustic modes factory-set; sadly you can't change it manually thanks to a patent dispute over AAM), but I still find them pretty quiet. Maybe you got unlucky with a slightly noiser than average drive (you do tend to get some variation in manufacture) and a case which happens to transmit the vibrations more than most.
"That's usually been the single source of the most noise in my systems. HD manufacturers need to make quieter drives."
Most drives I've bought in the past few years have been practically silent. The loudest's probably the ball bearing Maxtor D740X, and even that's drowned out by even the slightest of fan noise.
"I used to be a hardcore Maxtor fanatic, but with three out of four drive failures (250 gig drives) in only two months, I am hunting for a new manufacturer."
While this sounds like a PSU issue, Maxtor do seem to have more problems than average atm (latest thing seems to be firmware problems). I generally avoid them these days, even if I have little more than (lots of) anecdotal evidence to suggest they're not the best of choices.
"it only took one "DeathStar" drive from IBM to convince me that they made shitty drives"
They had a bad batch; before then they had a good reputation which wasn't entirely unearned.. the real problem was how badly they handled the obvious design flaws they had with those drives (which I believe were mostly caused by running them too hot; a problem drives like the Quantum Fireball AS had too).
"I can't stand Western Digital since I had three drives in a row fail"
Wait.. that's 7 failures? Unless you're running hundreds of these drives that's a bit extensive. I've said it before, and I'll say it again: Tips on running healthy HD's:
Buy a good PSU. This means the likes of Antec, Enermax, Sparkle, etc, with nice heavy heatsinks and words like "Active PFC" on the packaging; not the $25 550W dual-fan thing made by $random_cheapo_vendor.
Keep your drives cool. A big, slow fan providing a bit of airflow could easily add years to the lifespan of your drives, but failing that just make sure they're well ventilated and not even approaching 50c.
Buy them from reputable dealers using reputable delivery companies. You might save $5 but it's a bit of a false economy if they're going to batter the drives around during shipping.
Treat them with respect; those impressive sounding maximum shock values really aren't.
Run a SMART monitoring tool and keep an eye on things like reallocated sector counts. Most vendors will happily replace any drive which is reallocating sectors, since that's a common sign of impending doom.
"So that leaves Seagate which I haven't had any experience with so far"
They're the only (S)ATA drives I bother with these days. They're quiet, fairly fast, come with 5 year warranties and seem among the most reliable.
Huh? I'm talking about MySQL on FreeBSD and you take it as an attack on FreeBSD's general performance? You've completely missed my point here; FreeBSD's great for most use, it just happens that MySQL on Linux performs much better than MySQL on FreeBSD (and just about everything else it seems).
This isn't an OS advocacy issue (I vastly prefer FreeBSD to Linux), this is a real problem for certain heavy MySQL users. I'm not on the "FreeBSD suxx Linux r0xx!!" side here, I'm on the "Someone please track down and remove the bottleneck(s) so we can depenguinate the few Linux servers we've been forced to migrate to" side. Sheesh.
Let's see, say I want to buy the latest Chemical Bros album. 11 tracks.
iTunes: 79p per song = £8.69. I get a set of 128kbps rights-limited AAC's I can play in exactly: iTunes and on an iPod. If I like I can (badly; no offset correction) burn a copy to CD and then rip it, but to avoid quality loss I need to store it losslessly.
Amazon: £8.49 (plus postage, which is free over £20). I get a physical CD and case with a lossless copy of my music, and I get to rip it to any format I care to name and play it in pretty much anything I like. Sadly null and void if it's copy "protected", since it's effectively pre-scratched (they mess up error correction).
Magnatune (if they did it): ~£4.24 (~$8). I get any format I like, either directly or by transcoding from lossless copies. For double that I get a physical CD, and also get to download it.
One of these rocks and has been where I've bought most of my recent music. One of these are reasonable and I've bought a few things from them. One of them is a complete joke, and will never get my custom. I'll let you work out which ones.
Let me guess, you don't follow any FreeBSD mailing lists, you've never done any MySQL benchmarks yourself, you've never spent countless hours tweaking everying imaginable to try to improve MySQL's performance under FreeBSD, and you've never had to manage a heavily loaded production MySQL cluster in a primarily FreeBSD environment?
These benchmarks are actually pretty close to what you get no matter how well you tune things. FreeBSD consistantly performs a lot poorer than Linux for some reason, although they're slowly improving.
Disk IO is irrelevent; you see this even on on read-only tests against completely cached tables. Believe me, I wish to hell it wasn't.
Works for me, although it locked up for about 15s after it loaded and I scrolled a bit, and RMB keeps bringing up Firefox's context menu as well as moving the map. It also brought up the Java console for some reason; maybe I turned it on ages ago and forgot about it.
Opera worked a lot better.
You do have a fairly recent version of Sun's JVM installed, right? Or are you trying to use the Microsoft JVM?
Slashdot Mirror
"It would be nice if the various linux filesystem drivers could have a mount option that spread out writes"
Flash drives already do load leveling in hardware; they are after all, usually used with FAT.
For the few cases where you need to do it yourself, that's what JFFS2 is for.
"The default install of XP refuses to boot off of my >137 GB"
Burn a slipstreamed image of XP+SP2.
"I shouldn't have to regularly reinstall everything due to a flakey registry to begin with"
Um, no, you shouldn't. What the hell are you doing, blasting your DIMMs with gamma rays or something?
"ultimatley, that's what's keeping me from going Linux as well. I know that getting the software up and runing on my machine will be just as bad if not worse than reinstalling XP. At least with XP there are drivers to be found"
I'd rather reinstall my FreeBSD workstation than my XP desktop. Driver wise a GENERIC kernel generally does better than XP in my experience, and reinstalling my apps is just a portinstall away. Better than grabbing DOpus, DigiGuide, Vim, PenguiNet, Opera, Firefox, daemon-tools, nVidia drivers, sound card drivers, cygwin, MPC, VLC, NBPro, ffdshow, Samurize, SpeedFan, Locate32, Nero, Azureus, TweakUI, Synergy, etc, all from different places. The difficulty's mostly reduced from actually going and finding those apps to just making a list. *shrug*, YMMV, swings+roundabouts, etc.
"sorry? OpenBSD wants GUI management information? Am I missing something here?"
RAID management involves things like (re)building arrays, adding spares, etc, which requires access to the management interface. Ignore the word "GUI" here; under Linux you access it using a tool called aaccli; under FreeBSD you can use it under Linux emulation (kinda).
Open source drivers are available and have been for years (if not always in a very usable state; Linux aacraid's only recently stopped being marked EXPERIMENTAL), but they don't include management tools; without them you're basically forced to talk to the card using its BIOS at POST, which is kinda suboptimal if you're trying to use hardware RAID to make a system more reliable.
Ironically AAC's support under FreeBSD at least has been superior to Linux's for quite a while (not so much in the past few months, but certainly for the past couple of years before). We originally bought our cards to run under FreeBSD, and had significant problems migrating to Linux where the aacraid driver liked to fall over every few weeks.
I'd rather use software RAID now. Closed source management tools and unreliable software, hardware and firmware are not things I want anywhere fucking near my data storage subsystems.
"you never hear *nix overclockers complain that their OS crashes all the time do you?"
Yes? Although more normally it's things like gcc throwing tonnes of SEGV's and the odd bit of filesystem corruption. Maybe Windows just has more assert()s.
> if(isset($_SERVER["HTTP_ACCEPT"]) && stristr($_SERVER["HTTP_ACCEPT"],"application/xhtml +xml"))
;)
This is a common mistake. You really should parse out each entry and select based on q-values. What happens when the client says application/xhtml+xml;q=0 ("XHTML is unacceptable"), or even text/html;q=1.0, application/xhtml+xml;q=0.6 ("I prefer HTML, but I can handle XHTML to some extent")? Or is it ok to ignore one spec if you kinda follow another better?
Oh, and don't forget to switch your <link rel="stylesheet"> bits with <?xml-stylesheet ?> when you're serving as XML.
Personally, I tend to just stick with HTML 4.01. If I'm going to use XHTML, I like to use an XSLT to transform it to HTML 4.01 for naughty little browsers, which also handily enforces well-formedness in a way that allows for degredation to text/html-served broken XHTML instead of a browser-level parse error.
It's a preview for a reason. And what higher standard are you holding Opera to? Firefox has plenty of bugs and misfeatures of its own, and yes, they do tend to go unfixed for a depressingly long time (up to and including never).
Certainly in my experience as a web developer, Opera's tended to fare better than Gecko and KHTML-based browsers with regard to layout problems and incomplete standards support. Opera's hardly perfect, but it gets closer than most.
These are peaks too; you'll only be hitting them with your laptop plugged into the wall and running cpuburn, otherwise they'll be ticking over in low power mode.
Creative? Do they still convert to 48khz internally and break the PCI spec in interesting ways? :/
Samba VFS Module, as linked from FLAC website.
Sadly they don't seem to have gone very far with their lossless service; most of their music is still transcoded from ~384kbps (free-format) MP3's. I've bought a few albums from them as FLAC and been very happy to be able to do so, but most of their selection still doesn't meet my requirements for "good quality" (yeah, so I'm a bit anal and probably would never notice the difference, but that's not the only reason I want lossless).
I still vastly prefer Magnatune; if iTunes were like that, I'd probably be bankrupt.
"You cannot accelerate networking very much because the problem is highly serial."
It's serial at line-level (usually), but that doesn't mean you can't process frames/packets in parallel. At the very least you can offload things like checksums, and process packets for different connections in parallel; you can even do simple things like aligning packet payloads on page boundaries so the OS doesn't have to copy them about needlessly (zero-copy is good when you're talking 100's of MB/s, no?). On top of that, I'm pretty sure you can process multiple packets for a single connection in parallel since you can fill different bits of your socket window from multiple packets at once.
"You have all that CPU for free;"
Er, no, you don't; you just said it's been paid for, it's certainly not free. In the case of a heavily loaded server, you don't want half your CPU doing little more than copying memory between the network and your application, doing tedious packet reassembly, calculating checksums and servicing more interrupts than necessary; you want it running your application code to generate that data.
"Much like RAID5 however, there is a space sacrifice for this extra parity layer."
But it's settable; so if you want to be able to recover fully from losing/corrupting 20% of your backup you just set it to 20% of your backup size, and if you only care about a few minor bit errors or so, you can drop it to a couple of percent or less.
Be nice if vendors provided PAR2's for their ISO/DVD images/anything else big; it sucks when you find the MD5 of your download doesn't match the one they provide (or that 400MB setup.exe throws a checksum mismatch and refuses to run), and you know it's probably just a single bit flipped somewhere but can't do much beyond redownloading the entire thing. rsync helps, of course, but that's a *lot* heavier on the server both in resource use and administration cost.
Funny; the two Maxtor DiamondMax 9's my friend just bought simply don't work properly with the Intel SATA chipset he wanted to use them with. They're fine on the Promise/Via controllers on his desktop, but on the Intel and libata they throw IO errors after about 50 sectors. The two Seagates he bought with them are fine...
A bit of research suggests these drives are prone to firmware bugs and outright failure, but of course Maxtor's site doesn't appear to mention any updates for them.. bah.
The SATA ones are a little buzzier than the PATA ones (they have different acoustic modes factory-set; sadly you can't change it manually thanks to a patent dispute over AAM), but I still find them pretty quiet. Maybe you got unlucky with a slightly noiser than average drive (you do tend to get some variation in manufacture) and a case which happens to transmit the vibrations more than most.
Most drives I've bought in the past few years have been practically silent. The loudest's probably the ball bearing Maxtor D740X, and even that's drowned out by even the slightest of fan noise.
"I used to be a hardcore Maxtor fanatic, but with three out of four drive failures (250 gig drives) in only two months, I am hunting for a new manufacturer."
While this sounds like a PSU issue, Maxtor do seem to have more problems than average atm (latest thing seems to be firmware problems). I generally avoid them these days, even if I have little more than (lots of) anecdotal evidence to suggest they're not the best of choices.
"it only took one "DeathStar" drive from IBM to convince me that they made shitty drives"
They had a bad batch; before then they had a good reputation which wasn't entirely unearned.. the real problem was how badly they handled the obvious design flaws they had with those drives (which I believe were mostly caused by running them too hot; a problem drives like the Quantum Fireball AS had too).
"I can't stand Western Digital since I had three drives in a row fail"
Wait.. that's 7 failures? Unless you're running hundreds of these drives that's a bit extensive. I've said it before, and I'll say it again: Tips on running healthy HD's:
"So that leaves Seagate which I haven't had any experience with so far"
They're the only (S)ATA drives I bother with these days. They're quiet, fairly fast, come with 5 year warranties and seem among the most reliable.
Huh? I'm talking about MySQL on FreeBSD and you take it as an attack on FreeBSD's general performance? You've completely missed my point here; FreeBSD's great for most use, it just happens that MySQL on Linux performs much better than MySQL on FreeBSD (and just about everything else it seems).
This isn't an OS advocacy issue (I vastly prefer FreeBSD to Linux), this is a real problem for certain heavy MySQL users. I'm not on the "FreeBSD suxx Linux r0xx!!" side here, I'm on the "Someone please track down and remove the bottleneck(s) so we can depenguinate the few Linux servers we've been forced to migrate to" side. Sheesh.
Let's see, say I want to buy the latest Chemical Bros album. 11 tracks.
iTunes: 79p per song = £8.69. I get a set of 128kbps rights-limited AAC's I can play in exactly: iTunes and on an iPod. If I like I can (badly; no offset correction) burn a copy to CD and then rip it, but to avoid quality loss I need to store it losslessly.
Amazon: £8.49 (plus postage, which is free over £20). I get a physical CD and case with a lossless copy of my music, and I get to rip it to any format I care to name and play it in pretty much anything I like. Sadly null and void if it's copy "protected", since it's effectively pre-scratched (they mess up error correction).
Magnatune (if they did it): ~£4.24 (~$8). I get any format I like, either directly or by transcoding from lossless copies. For double that I get a physical CD, and also get to download it.
One of these rocks and has been where I've bought most of my recent music. One of these are reasonable and I've bought a few things from them. One of them is a complete joke, and will never get my custom. I'll let you work out which ones.
Let me guess, you don't follow any FreeBSD mailing lists, you've never done any MySQL benchmarks yourself, you've never spent countless hours tweaking everying imaginable to try to improve MySQL's performance under FreeBSD, and you've never had to manage a heavily loaded production MySQL cluster in a primarily FreeBSD environment?
These benchmarks are actually pretty close to what you get no matter how well you tune things. FreeBSD consistantly performs a lot poorer than Linux for some reason, although they're slowly improving.
Disk IO is irrelevent; you see this even on on read-only tests against completely cached tables. Believe me, I wish to hell it wasn't.
"And crashes Firefox for me (v1.0, Win XP Pro)."
Works for me, although it locked up for about 15s after it loaded and I scrolled a bit, and RMB keeps bringing up Firefox's context menu as well as moving the map. It also brought up the Java console for some reason; maybe I turned it on ages ago and forgot about it.
Opera worked a lot better.
You do have a fairly recent version of Sun's JVM installed, right? Or are you trying to use the Microsoft JVM?
Why they don't do this by default is beyond me, why bother making a dynamically loading plugin system when you just load everything by default anyway?