Ext3 writes out data before metadata is written (at lest in the default mode data=ordered), this there is no window opportunity where a crash could cause a data loss on ext3. On ext4 there is a 60 second window of opportunity. Or was, before this bug was fixed by patches pending for 2.6.30.
The new data is NOT important, it can be thrown away in a crash an no one will complain. The problem is that ext4 managed to destroy data that was already on the disk. That is unacceptable.
ext3 is also delaying writes. The bug is that ext4 is not delaying renames to happen after writes. Instead renames happen immediately, and guess what, they spin your hard drive up, then you get to wait 60 second until real data starts to be written.
Oh and if you lose power or crash during these 60 seconds, you loose all data - new and old. Oh and you common desktops programs do that cycle several times a minute.
Amost correct, but what actually happens with ext3 is the following:
* 1. Create config.new. (Should be empty, because it's new)
* 2. Write the new contents into config.new (cached)
* 3. Move config.new onto config (cached)
(time passes)
* 3b. Filesystem decides that it is time to commit cache to disk and tries to commit metadata first. All commits are written to a journal
* 2b. Metadata commit is determined to be dependant on file data, so file data is written first.
* 3c. Metadata is written do disk.
If a crash happens at any point before 3c, after crash you get the old file, if after 3c, you get the new file.
The right thing is to code a reliable filesystem, despite POSIX not demanding such feat. POSIX is written in the 80s when system level code was crap. Now we can do better than that.
fsync() is no an answer - we do not care if the old data or the new data is saved after the crash. Ext4 looses both. If we start using fsync() that will hurt performance by a ton, cripple write caching, destroy laptop battery life, wear out SSDs much faster than really necessary and cause a bunch of other side effects without any actual gain except hiding this filesystem regression.
It is a filesystem bug - regression with a mayor data loss in common usage scenarions.
A filesystem could erase files from disk every time read() is done and still be perfectly POSIX compilant (if they would put it back on a fsync() call), however that would also be retarded and an outright disregard to valuable user data.
Same here - the bug shows utter disregard to user data. POSIX compilance here is just as irrelevant as whether the code is indented according to C coding guidelines or not. It is still a regression from ext3 and a data loss under common usage scenarios.
A few percent performance difference will be easily wiped away when the filesystem erases an important file that one time a year when a snowstorm knocks your power out.
1) Modern filesystems are expected behave better than POSIX demands.
2) POSIX does not cover what should happen in a system crash at all.
3) The issue is not about saving data, but the atomicity of updates so that either the new data or the old data would be saved at all times.
4) fsync is not a solution, because ir forces the operation to complete *now*, which is counterproductive to write performance, cache coherence, laptop battery life, excessive SSD wear and a bunch of other reasons.
We don't need reliable data-on-disk-now, we need reliable old-or-new data without using a sledgehammer of fsync.
No, we will install a box on a back wall of my house overlooking the backyard and with a LoS t most windows and then turn it on.
After that we will continually upgrade the software until I can keep my windows open in the hot summer nights without becoming big and red from the damn bites.
After that we sell millions of those compact boxes to other people in the US with an option to send another of those boxes over to a malaria infested village in the 3rd world.
Shape, not throttle. If done correctly shaping is what makes a difference between a good ISP and a great ISP. It is not a problem to detect P2P traffic and shape it to a lower priority, provided that you shape important traffic as high priority - ACK's, Skype voice, game traffic (WoW, CS,...), first 100k of any HTTP or HTTPS connection, SSH,...
As a power user it is not that critically important that my torrents only come at 16kb/s during the day if my web, games and IM apps are snappy, but I would like to have the torrents saturate the pipe during off-peak.
Also, hard caps are overrated - you don't pay per Gb, why should we? Just prioritise traffic correctly and everyone will be happy.
And this time it is no meme, but a real question. What good are slide-out screens and fancy fingerprint readers if they are based on such obscene hardware hacks that a normal operation system would be unable to use it all.
That is something reviews would actually be useful for.
Earth has not been 'far warmer'. See the hockey stick.
Cold in DC and in Europe for that matter is due to slowing of the Golfstream and masses on polar region air coming down, both of those phenomenons are activated by global warming.
It could be that the natural negative feedback loop for the global warming is the formation of an Ice age cowering Europe and most of the USA under miles of ice. That might balance things out. In a couple thousand years.
So... 10 billion people, frozen Europe, Canada, USA, Russia, scorched deserts in most of Afrika... where will we all go to live? Will everyone migrate peacefully? How many billion people will be killed in wars to control few strips of land that are still fertile? And how many billions will die of starvation because they did not have the military power to get those lands?
The funny thing is that the actul bug is an urban myth. People claim that once your hard drive reaches 300k parks it will fail. Note that at the start they were claiming the number to be 100k and now are claiming 600k due to the simple fact that a huge number of people showed up with the number being well over million on perfectly functioning drives.
The drives parked heads when not in use, sometimes, several times a minute, some of them clicked when they did so. It is a feature that reduces power use and protects the hard drive from sudden movement and impacts. It is NOT a bug.
All the claims that it will make you hard drive fail in a year are false and are made by people that have no a slightest clue of hard drive design.
I, however, am more worried about the wear on the cable than on the bristles. Do that experiment with a broomstick and then look at the surface of the broom handle.
Easy solution - go to EU. The patent as described is a pure software patent and would be invalid outside US (and Japan). Way to squash innovation, US.:P
Badly wrong, sorry. There actually is no specific word to describe a software distribution model where source code is available to customers on a non-OSI-approved licence. Founders of OSI created the term 'open source' and they defined it with rules taken from DFSG. And there is a lot of different meanings of the word 'free', but in educated circles 'free software' = FLOSS = software with an OSI-approved licence or a DFSG compilant licence. Free as in 'free beer' is another example of a different meaning of 'free' that does not apply.
If you transport a book in a paper wrapping, it gets wet and an imprint of the cover art is left on the wrapping paper. That is the same level of distribution ingrained in the delivery of the work to you as the redistribution of file parts is in a P2P network file delivery process.
IMHO the theory that you are distributing when you download a file via P2P was already defeated in a court in US in one of those RIAA court cases, but IANAL.
True, provided that noone gets physically harmed or killed in the process, which was also in his ideology. That book helped me develop critical thinking when I was young.
You just shot your argument in the face. Netbooks are good enough, while a MacBook Pro really is great.
As a free software developer I must say that very often free software project lack the focus and leadership to really polish their product to greatness like Apple does. We do have a lot to learn from them. There is no shame in admitting that someone does something better than you do and then earning from that.
And the other side of the coin is that IT is not a producing industry. IT merely allows other industries to produce their goods and services in a more efficient fashion. From this you can clearly see that the real source of money for IT is serving other industries as custom solutions.
Commodity market can go to 0 without a significant impact on global IT economy, because even now 9 out of 10 programmers work for non-IT companies. If your company is not selling software, then raise of free software is only to your benefit.
And even more importantly - if something does go wrong, the dying astronaut might be able to say what it was before dying. That chance alone is move valuable than any controller discomfort.
For preventative tasks (which should be where a good sysadmin spends most of his time) I have two words - "risk assessment". Make a table of risks that you are preventing, probability of these risks occurring if your preventative actions are not taken and cost if the worst possible outcome strikes. Multiply and sum as appropriate and you will get a very nice looking figure of prevented cost.
IT industry is a tiny industry compared to all the others. While free software might be slightly worse for the IT industry (which is quite debatable), it is unquestionably for the better for all the other industries.
If your company is selling cars and one of your IT guys writes a patch that improves the power efficiency of your servers by 1% it is good for you. However, it will not pay for you to banch out into a software business to try to sell that patch to others. It is cheaper to give it away for free, thus your contribution creates value for others as well. The funny thing is that when other companies do the same, your company benefits as well. This creates a natural positive feedback of innovation that does not need money to increase exponentially. It does require that your company makes money at your primary business so you can pay your employees, however.
Slashdot Mirror
2. This bug does NOT exist in ext3.
Ext3 writes out data before metadata is written (at lest in the default mode data=ordered), this there is no window opportunity where a crash could cause a data loss on ext3. On ext4 there is a 60 second window of opportunity. Or was, before this bug was fixed by patches pending for 2.6.30.
The new data is NOT important, it can be thrown away in a crash an no one will complain. The problem is that ext4 managed to destroy data that was already on the disk. That is unacceptable.
ext3 is also delaying writes. The bug is that ext4 is not delaying renames to happen after writes. Instead renames happen immediately, and guess what, they spin your hard drive up, then you get to wait 60 second until real data starts to be written.
Oh and if you lose power or crash during these 60 seconds, you loose all data - new and old. Oh and you common desktops programs do that cycle several times a minute.
Have fun.
Amost correct, but what actually happens with ext3 is the following:
* 1. Create config.new. (Should be empty, because it's new)
* 2. Write the new contents into config.new (cached)
* 3. Move config.new onto config (cached)
(time passes)
* 3b. Filesystem decides that it is time to commit cache to disk and tries to commit metadata first. All commits are written to a journal
* 2b. Metadata commit is determined to be dependant on file data, so file data is written first.
* 3c. Metadata is written do disk.
If a crash happens at any point before 3c, after crash you get the old file, if after 3c, you get the new file.
The right thing is to code a reliable filesystem, despite POSIX not demanding such feat. POSIX is written in the 80s when system level code was crap. Now we can do better than that.
fsync() is no an answer - we do not care if the old data or the new data is saved after the crash. Ext4 looses both. If we start using fsync() that will hurt performance by a ton, cripple write caching, destroy laptop battery life, wear out SSDs much faster than really necessary and cause a bunch of other side effects without any actual gain except hiding this filesystem regression.
It is a filesystem bug - regression with a mayor data loss in common usage scenarions.
A filesystem could erase files from disk every time read() is done and still be perfectly POSIX compilant (if they would put it back on a fsync() call), however that would also be retarded and an outright disregard to valuable user data.
Same here - the bug shows utter disregard to user data. POSIX compilance here is just as irrelevant as whether the code is indented according to C coding guidelines or not. It is still a regression from ext3 and a data loss under common usage scenarios.
A few percent performance difference will be easily wiped away when the filesystem erases an important file that one time a year when a snowstorm knocks your power out.
In this case, all of XFS experience on the topic applies to Ext4 perfectly.
1) Modern filesystems are expected behave better than POSIX demands.
2) POSIX does not cover what should happen in a system crash at all.
3) The issue is not about saving data, but the atomicity of updates so that either the new data or the old data would be saved at all times.
4) fsync is not a solution, because ir forces the operation to complete *now*, which is counterproductive to write performance, cache coherence, laptop battery life, excessive SSD wear and a bunch of other reasons.
We don't need reliable data-on-disk-now, we need reliable old-or-new data without using a sledgehammer of fsync.
You have a separate partition for /root ? How large can the home folder of the root user be?
No, we will install a box on a back wall of my house overlooking the backyard and with a LoS t most windows and then turn it on.
After that we will continually upgrade the software until I can keep my windows open in the hot summer nights without becoming big and red from the damn bites.
After that we sell millions of those compact boxes to other people in the US with an option to send another of those boxes over to a malaria infested village in the 3rd world.
???
Profit!
Shape, not throttle. If done correctly shaping is what makes a difference between a good ISP and a great ISP. It is not a problem to detect P2P traffic and shape it to a lower priority, provided that you shape important traffic as high priority - ACK's, Skype voice, game traffic (WoW, CS, ...), first 100k of any HTTP or HTTPS connection, SSH, ...
As a power user it is not that critically important that my torrents only come at 16kb/s during the day if my web, games and IM apps are snappy, but I would like to have the torrents saturate the pipe during off-peak.
Also, hard caps are overrated - you don't pay per Gb, why should we? Just prioritise traffic correctly and everyone will be happy.
http://www.streamingcolour.com.nyud.net/blog/2009/03/09/the-numbers-post-aka-brutal-honesty/
And this time it is no meme, but a real question. What good are slide-out screens and fancy fingerprint readers if they are based on such obscene hardware hacks that a normal operation system would be unable to use it all.
That is something reviews would actually be useful for.
Earth has not been 'far warmer'. See the hockey stick.
Cold in DC and in Europe for that matter is due to slowing of the Golfstream and masses on polar region air coming down, both of those phenomenons are activated by global warming.
It could be that the natural negative feedback loop for the global warming is the formation of an Ice age cowering Europe and most of the USA under miles of ice. That might balance things out. In a couple thousand years.
So ... 10 billion people, frozen Europe, Canada, USA, Russia, scorched deserts in most of Afrika ... where will we all go to live? Will everyone migrate peacefully? How many billion people will be killed in wars to control few strips of land that are still fertile? And how many billions will die of starvation because they did not have the military power to get those lands?
Not so fun.
The funny thing is that the actul bug is an urban myth. People claim that once your hard drive reaches 300k parks it will fail. Note that at the start they were claiming the number to be 100k and now are claiming 600k due to the simple fact that a huge number of people showed up with the number being well over million on perfectly functioning drives.
The drives parked heads when not in use, sometimes, several times a minute, some of them clicked when they did so. It is a feature that reduces power use and protects the hard drive from sudden movement and impacts. It is NOT a bug.
All the claims that it will make you hard drive fail in a year are false and are made by people that have no a slightest clue of hard drive design.
I, however, am more worried about the wear on the cable than on the bristles. Do that experiment with a broomstick and then look at the surface of the broom handle.
Easy solution - go to EU. The patent as described is a pure software patent and would be invalid outside US (and Japan). Way to squash innovation, US. :P
Badly wrong, sorry. There actually is no specific word to describe a software distribution model where source code is available to customers on a non-OSI-approved licence. Founders of OSI created the term 'open source' and they defined it with rules taken from DFSG.
And there is a lot of different meanings of the word 'free', but in educated circles 'free software' = FLOSS = software with an OSI-approved licence or a DFSG compilant licence. Free as in 'free beer' is another example of a different meaning of 'free' that does not apply.
If you transport a book in a paper wrapping, it gets wet and an imprint of the cover art is left on the wrapping paper. That is the same level of distribution ingrained in the delivery of the work to you as the redistribution of file parts is in a P2P network file delivery process.
IMHO the theory that you are distributing when you download a file via P2P was already defeated in a court in US in one of those RIAA court cases, but IANAL.
True, provided that noone gets physically harmed or killed in the process, which was also in his ideology. That book helped me develop critical thinking when I was young.
You just shot your argument in the face. Netbooks are good enough, while a MacBook Pro really is great.
As a free software developer I must say that very often free software project lack the focus and leadership to really polish their product to greatness like Apple does. We do have a lot to learn from them. There is no shame in admitting that someone does something better than you do and then earning from that.
And the other side of the coin is that IT is not a producing industry. IT merely allows other industries to produce their goods and services in a more efficient fashion. From this you can clearly see that the real source of money for IT is serving other industries as custom solutions.
Commodity market can go to 0 without a significant impact on global IT economy, because even now 9 out of 10 programmers work for non-IT companies. If your company is not selling software, then raise of free software is only to your benefit.
And even more importantly - if something does go wrong, the dying astronaut might be able to say what it was before dying. That chance alone is move valuable than any controller discomfort.
For preventative tasks (which should be where a good sysadmin spends most of his time) I have two words - "risk assessment". Make a table of risks that you are preventing, probability of these risks occurring if your preventative actions are not taken and cost if the worst possible outcome strikes. Multiply and sum as appropriate and you will get a very nice looking figure of prevented cost.
IT industry is a tiny industry compared to all the others. While free software might be slightly worse for the IT industry (which is quite debatable), it is unquestionably for the better for all the other industries.
If your company is selling cars and one of your IT guys writes a patch that improves the power efficiency of your servers by 1% it is good for you. However, it will not pay for you to banch out into a software business to try to sell that patch to others. It is cheaper to give it away for free, thus your contribution creates value for others as well. The funny thing is that when other companies do the same, your company benefits as well. This creates a natural positive feedback of innovation that does not need money to increase exponentially. It does require that your company makes money at your primary business so you can pay your employees, however.