I think for very cheap machines, if you take the cost difference between an AMD CPU + motherboard vs. Intel CPU + motherboard and put that cost difference into an SSD while the Intel box still has a traditional hard drive, then AMD is a good value.
And in fact, that's what I did with my wife's most recent computer. AMD A8-7600 + 12GB of RAM + 120GB SSD. Extremely cheap and it can still play Minecraft and Call of Duty: Black Ops 2 for my sons.
But if you were going to get an SSD anyway, plus 6+GB of RAM (more for a power user or developer or someone doing video editing or virtualization), then I agree with you. Paying the extra $100 to go from an $80 AMD "APU" to a $120 Intel i3-4160 and compatible motherboard will pay off in spades. Even the $70 Pentium dual core 50 watt G3258 kills any AMD processor this side of the overclocked 220 watt FX series chips for single-threaded performance.
It's also real life imitating art, because the protagonist of Charles Stross' novel Accelerando ( https://en.wikipedia.org/wiki/... ) uses this technique and becomes persona non-grata in the US. The book came out in 2005.
Good point. I did mean to include that under the inferior and inadequate solutions options, but of course it does complicate things because finding a technical solution that works that also has your preferred combination of license and contributor license agreement is much harder than just creating a technical solution that works.
I have a 1055T (not trying to one-up you) and a Radeon 5770. It worked fine for Windows 7 when I had it, but I had bought Win7 retail so when my wife's Windows XP install croaked I bought an SSD for her machine and put the Win7 install on that, and since then I've run Linux full time.
The Anonymous Coward a few posts back pointed that out to me already. I forgot the third possibility - that my fourteenth standard is at best no better than the best of the thirteen others and I have just made the situation worse. You and he or she are correct about that, and I'm sorry I didn't include it in my list of options.
Good luck. If you're familiar with Red Hat and CentOS, then Fedora may work too. I have Ubuntu on one machine and Fedora on another. I like a lot of Linux distributions, but I try to use the most mainstream ones so that if a casual Linux user among my friends or family has a question, I have up to date experience with something close to their setup.
Systemd is more work for you, someone with a broad depth of expertise in SysV init and shell scripts. For someone new to both, systemd is less work and provides all of the same features and advantages. Lennart Poettering is not a Red Hat executive or a Fedora guidance committee chairman - he and his team got systemd adapted by Fedora, then OpenSUSE, then Arch, then Debian, then Ubuntu, then CentOS based only on its technical merits.
I have encountered all of that information too. However, the situation has been improving. I can't say how much of that improvement is due to AMD and how much is due to open source developers that are good at reverse engineering. http://www.phoronix.com/scan.p... Also, I had those Minecraft crashes on Ubuntu 14.04, 14.10, and 15.04 alpha - that article I linked has kernel 3.18 (same as Ubuntu 15.04 alpha), Mesa 10.5-devel (15.04 alpha has 10.4.something), and open source radeon driver 7.4.99 (15.04 has 7.4.0). Maybe if I compiled my own driver and mesa I would get better stability, but I can't be sure.
I'm an AMD fan from way back because of the monopoly tricks Intel pulled in the late 1990s early 2000s ( https://en.wikipedia.org/wiki/.... ) - I figure Intel has the money and resources to put towards open source today because of the advantage they unfairly gained due to tricks then.
Thanks for catching my failure at reading comprehension. In any event, that still doesn't match what I'm seeing on Fedora 21. "journalctl -r -o verbose | grep -i 'priority=3' " gives hits, as does grep for priority 2. I'm not getting any hits for 1 or 0, but presumably that's because nothing has gone catastrophically wrong.
1. systemctl-journal-gatewayd sends the digitally signed logs over https to another machine, same thing with all of the digital signing and security benefits.
2. Faster never hurts. There's no extra money made by waiting longer.
3. You wrote "A prod server should not be running piles of things". My point is with systemd you don't have to. I never said init did not also fit that criteria.
4. Sure. But it makes it easier. Just a few text file entries.
5. No, as far as I understand systemd socket activation the startup cost is incurred once on the first request, then after that the process keeps running. So best possible combination.
6. Again, less work to set all of this up. Just a text file.
7. Again, less work to get this feature. With systemd, you get this with "systemctl restart foo.service". No extra commands in every single init script for every single service to make sure all resources are closed down.
8. And you can keep writing them. systemd is compatible with init scripts. So what's the drawback?
That's bizarre. I'm not a sysadmin, I'm only running Linux on my work laptop and my personal desktop. For Fedora 21, if I do "journalctl -r -o verbose" I'm getting plenty of priority 6 and priority 7 messages in the output.
AMD graphics, yes. With the open source drivers. I understand the drivers have gotten better over time, but I guess they're not up to Windows or Linux nVidia proprietary driver standards.
Are you trolling? I've never heard that the systemd journal discards startup information. I'm seeing everything on my Fedoral install. dmesg looks normal too.
Besides, you can run rsyslog alongside the systemd journal.
s/we the users/we the people who want open source software to be more popular and are willing to work to that effect/
I didn't mean to imply users are expected or forced to contribute. I meant that fans - and I consider myself a fan - that want this to be more popular should contribute because it will help that goal. Happy now?
Sure it is useful on the server. In no particular order:
1. The journal digitally signs each entry with the entry contents and the hash of the previous entry, so that for an attacker to insert a spoofed entry or remove a valid one they have to alter the signature on every entry after that point or else the signature mismatch will be detected. And you can still also send logs to rsyslog if you want.
2. Faster boot time does matter for a server - when you need to move physical boxes, add hardware that requires a power cycle, and so forth less time to restart is helpful.
3. You don't need to run "piles of things" with systemd. It's modular, so you only need to compile in the services you want. http://freedesktop.org/wiki/So...
4. Systemd lets you set limits on resource usage by each service: memory limits, CPU limits, Block IO limits, etc... which is useful on the server.
5. On-demand socket-driven service start is useful, so ssh is available 100% of the time but sshd isn't actually consuming resources until it receives a connection attempt.
6. Per-service private/tmp directories, configurable read-only access to some directories, so a hacked service can't access information it should not or make invalid writes.
7. Because of the use of cgroups in the Linux kernel, when you halt a service you can be confident there are no uncleaned resources - threads, forked processes, file handles, etc... left in use.
8. systemd is compatible with services written in any programming language you want, including shell scripts, so you don't have to rewrite your custom SysV init service in systemd. Just spent a few hours in the documentation to make the text file you need, and you're set.
You're one of the people who didn't actually read any of the systemd documentation. Why should I waste my time responding to you? But because I'm stupid, here it goes:
1. FreeBSD, OS X, and Solaris have all moved on from SysV init. Do you think they did that because it was fun? Because they felt like it? Or maybe because there were design flaws and maintenance headaches in SysV init that needed to be addressed?
2. You can run the systemd journal alongside rsyslog. And the journalctl command for getting text output from the systemd journal takes all of twenty minutes to learn.
3. Systemd services are written in C by convention with text configuration files. But you can run any executable you want as a systemd service: shell scripts, C++, Perl, Java, Lisp, Ruby, COBOL.
I didn't say 10% were people with legitimate complaints, I said 10% were people who understood the tradeoffs. A tradeoff implies advantages as well as disadvantages. So it's not the end of the world.
Nope, because usually when you add a new drive you reinstall Linux. It's free. Anyone who understands enough to transfer the whole filesystem image from one disk to another disk is going to update/etc/fstab manually.
Of course I understand where you're coming from. I'm practical, too - I run mostly free (as in freedom) software, but I work on proprietary software to pay the mortgage.
But the thing is, by living that way you let Microsoft or Apple control aspects of your digital life that they have no right to touch. Consider that.
You admin Red Hat and Solaris. I'm sure the price premium of proprietary software also isn't a headache for you unless maybe you wanted a private copy of Oracle 11. But for most of the population, $100 for Windows whatever or $400+ for a new machine or $150 for a professional reinstall after they get hit with malware is all a very big deal. I was having budget problems and was making less than half my current income when I got interested in Linux. I could afford to pay the Microsoft or Apple tax now - but thanks to the skills I developed then, I don't need to. I donate some of the difference to Debian and keep the rest.
Every dollar we throw at Microsoft and Apple - and Google, Oracle, etc... - strengthens proprietary software, strengthens the hold of Digital Rights Management, makes it easier for the government to snoop on all of us as a matter of routine, and makes it harder for poor people to get a good computing experience. Every dollar we put into open source strengthens open source software, weakens DRM, weakens privacy invasions, and makes it easier for regular people to have a good technology experience without spending more than they should budget to get it.
I'm not an expert in X, so I'll have to lookup "grab_break" and xkb and screen locking screensavers.
With respect to the systemd binary journal, a design document is here: https://docs.google.com/docume... Each entry is digitally signed with the hash of the previous entry. So any attacker that gets root can rewrite an entry, but in order to make the digital signatures pass verification he's got to rewrite the digital signature on the modified log entry and on every log entry from that point forward in time - feasible, but a lot more work than just modifying a text file and then changing the timestamp in a traditional log. and you can use rsyslog alongside journald.
Slashdot Mirror
I think for very cheap machines, if you take the cost difference between an AMD CPU + motherboard vs. Intel CPU + motherboard and put that cost difference into an SSD while the Intel box still has a traditional hard drive, then AMD is a good value.
And in fact, that's what I did with my wife's most recent computer. AMD A8-7600 + 12GB of RAM + 120GB SSD. Extremely cheap and it can still play Minecraft and Call of Duty: Black Ops 2 for my sons.
But if you were going to get an SSD anyway, plus 6+GB of RAM (more for a power user or developer or someone doing video editing or virtualization), then I agree with you. Paying the extra $100 to go from an $80 AMD "APU" to a $120 Intel i3-4160 and compatible motherboard will pay off in spades. Even the $70 Pentium dual core 50 watt G3258 kills any AMD processor this side of the overclocked 220 watt FX series chips for single-threaded performance.
It's also real life imitating art, because the protagonist of Charles Stross' novel Accelerando ( https://en.wikipedia.org/wiki/... ) uses this technique and becomes persona non-grata in the US. The book came out in 2005.
Good point. I did mean to include that under the inferior and inadequate solutions options, but of course it does complicate things because finding a technical solution that works that also has your preferred combination of license and contributor license agreement is much harder than just creating a technical solution that works.
I have a 1055T (not trying to one-up you) and a Radeon 5770. It worked fine for Windows 7 when I had it, but I had bought Win7 retail so when my wife's Windows XP install croaked I bought an SSD for her machine and put the Win7 install on that, and since then I've run Linux full time.
Try mosh, if you haven't already. :)
No mods, openjdk. Someone else pointed out elsewhere that the problem may be the open source AMD drivers.
The Anonymous Coward a few posts back pointed that out to me already. I forgot the third possibility - that my fourteenth standard is at best no better than the best of the thirteen others and I have just made the situation worse. You and he or she are correct about that, and I'm sorry I didn't include it in my list of options.
Good luck. If you're familiar with Red Hat and CentOS, then Fedora may work too. I have Ubuntu on one machine and Fedora on another. I like a lot of Linux distributions, but I try to use the most mainstream ones so that if a casual Linux user among my friends or family has a question, I have up to date experience with something close to their setup.
You can also do that with the systemd journal. Where's the problem?
Systemd is more work for you, someone with a broad depth of expertise in SysV init and shell scripts. For someone new to both, systemd is less work and provides all of the same features and advantages. Lennart Poettering is not a Red Hat executive or a Fedora guidance committee chairman - he and his team got systemd adapted by Fedora, then OpenSUSE, then Arch, then Debian, then Ubuntu, then CentOS based only on its technical merits.
I have encountered all of that information too. However, the situation has been improving. I can't say how much of that improvement is due to AMD and how much is due to open source developers that are good at reverse engineering. http://www.phoronix.com/scan.p... Also, I had those Minecraft crashes on Ubuntu 14.04, 14.10, and 15.04 alpha - that article I linked has kernel 3.18 (same as Ubuntu 15.04 alpha), Mesa 10.5-devel (15.04 alpha has 10.4.something), and open source radeon driver 7.4.99 (15.04 has 7.4.0). Maybe if I compiled my own driver and mesa I would get better stability, but I can't be sure.
I'm an AMD fan from way back because of the monopoly tricks Intel pulled in the late 1990s early 2000s ( https://en.wikipedia.org/wiki/.... ) - I figure Intel has the money and resources to put towards open source today because of the advantage they unfairly gained due to tricks then.
Thanks for catching my failure at reading comprehension. In any event, that still doesn't match what I'm seeing on Fedora 21. "journalctl -r -o verbose | grep -i 'priority=3' " gives hits, as does grep for priority 2. I'm not getting any hits for 1 or 0, but presumably that's because nothing has gone catastrophically wrong.
1. systemctl-journal-gatewayd sends the digitally signed logs over https to another machine, same thing with all of the digital signing and security benefits.
2. Faster never hurts. There's no extra money made by waiting longer.
3. You wrote "A prod server should not be running piles of things". My point is with systemd you don't have to. I never said init did not also fit that criteria.
4. Sure. But it makes it easier. Just a few text file entries.
5. No, as far as I understand systemd socket activation the startup cost is incurred once on the first request, then after that the process keeps running. So best possible combination.
6. Again, less work to set all of this up. Just a text file.
7. Again, less work to get this feature. With systemd, you get this with "systemctl restart foo.service". No extra commands in every single init script for every single service to make sure all resources are closed down.
8. And you can keep writing them. systemd is compatible with init scripts. So what's the drawback?
That's bizarre. I'm not a sysadmin, I'm only running Linux on my work laptop and my personal desktop. For Fedora 21, if I do "journalctl -r -o verbose" I'm getting plenty of priority 6 and priority 7 messages in the output.
Anybody else have access to Red Hat 7 to check?
AMD graphics, yes. With the open source drivers. I understand the drivers have gotten better over time, but I guess they're not up to Windows or Linux nVidia proprietary driver standards.
Are you trolling? I've never heard that the systemd journal discards startup information. I'm seeing everything on my Fedoral install. dmesg looks normal too.
Besides, you can run rsyslog alongside the systemd journal.
s/we the users/we the people who want open source software to be more popular and are willing to work to that effect/
I didn't mean to imply users are expected or forced to contribute. I meant that fans - and I consider myself a fan - that want this to be more popular should contribute because it will help that goal. Happy now?
Sure it is useful on the server. In no particular order:
/tmp directories, configurable read-only access to some directories, so a hacked service can't access information it should not or make invalid writes.
1. The journal digitally signs each entry with the entry contents and the hash of the previous entry, so that for an attacker to insert a spoofed entry or remove a valid one they have to alter the signature on every entry after that point or else the signature mismatch will be detected. And you can still also send logs to rsyslog if you want.
2. Faster boot time does matter for a server - when you need to move physical boxes, add hardware that requires a power cycle, and so forth less time to restart is helpful.
3. You don't need to run "piles of things" with systemd. It's modular, so you only need to compile in the services you want. http://freedesktop.org/wiki/So...
4. Systemd lets you set limits on resource usage by each service: memory limits, CPU limits, Block IO limits, etc... which is useful on the server.
5. On-demand socket-driven service start is useful, so ssh is available 100% of the time but sshd isn't actually consuming resources until it receives a connection attempt.
6. Per-service private
7. Because of the use of cgroups in the Linux kernel, when you halt a service you can be confident there are no uncleaned resources - threads, forked processes, file handles, etc... left in use.
8. systemd is compatible with services written in any programming language you want, including shell scripts, so you don't have to rewrite your custom SysV init service in systemd. Just spent a few hours in the documentation to make the text file you need, and you're set.
I'm running Ubuntu 15.04 alpha and Fedora 21, and /dev/hd?? and /dev/sd?? are still supported. UUIDs are just preferred.
You're one of the people who didn't actually read any of the systemd documentation. Why should I waste my time responding to you? But because I'm stupid, here it goes:
1. FreeBSD, OS X, and Solaris have all moved on from SysV init. Do you think they did that because it was fun? Because they felt like it? Or maybe because there were design flaws and maintenance headaches in SysV init that needed to be addressed?
2. You can run the systemd journal alongside rsyslog. And the journalctl command for getting text output from the systemd journal takes all of twenty minutes to learn.
3. Systemd services are written in C by convention with text configuration files. But you can run any executable you want as a systemd service: shell scripts, C++, Perl, Java, Lisp, Ruby, COBOL.
Come back and raise an objection after you RTFM.
I didn't say 10% were people with legitimate complaints, I said 10% were people who understood the tradeoffs. A tradeoff implies advantages as well as disadvantages. So it's not the end of the world.
Nope, because usually when you add a new drive you reinstall Linux. It's free. Anyone who understands enough to transfer the whole filesystem image from one disk to another disk is going to update /etc/fstab manually.
Of course I understand where you're coming from. I'm practical, too - I run mostly free (as in freedom) software, but I work on proprietary software to pay the mortgage.
But the thing is, by living that way you let Microsoft or Apple control aspects of your digital life that they have no right to touch. Consider that.
You admin Red Hat and Solaris. I'm sure the price premium of proprietary software also isn't a headache for you unless maybe you wanted a private copy of Oracle 11. But for most of the population, $100 for Windows whatever or $400+ for a new machine or $150 for a professional reinstall after they get hit with malware is all a very big deal. I was having budget problems and was making less than half my current income when I got interested in Linux. I could afford to pay the Microsoft or Apple tax now - but thanks to the skills I developed then, I don't need to. I donate some of the difference to Debian and keep the rest.
Every dollar we throw at Microsoft and Apple - and Google, Oracle, etc... - strengthens proprietary software, strengthens the hold of Digital Rights Management, makes it easier for the government to snoop on all of us as a matter of routine, and makes it harder for poor people to get a good computing experience. Every dollar we put into open source strengthens open source software, weakens DRM, weakens privacy invasions, and makes it easier for regular people to have a good technology experience without spending more than they should budget to get it.
I'm not an expert in X, so I'll have to lookup "grab_break" and xkb and screen locking screensavers.
With respect to the systemd binary journal, a design document is here: https://docs.google.com/docume... Each entry is digitally signed with the hash of the previous entry. So any attacker that gets root can rewrite an entry, but in order to make the digital signatures pass verification he's got to rewrite the digital signature on the modified log entry and on every log entry from that point forward in time - feasible, but a lot more work than just modifying a text file and then changing the timestamp in a traditional log. and you can use rsyslog alongside journald.
It was? When? I've been using since 2001.