True, but I was referring to a work desktop computer. I get to run desktop software (browser, office apps, terminals, e-mail, development IDE, etc) not encoding software or number-crunching. The only time it's not waiting for input is when the programs are loading, either themselves or data. About the most intensive data it gets to process (In the CPU) are encrypted e-mails or ssh sessions.
Thanks for your insightful and helpful remarks. Sarcasm aside, you don't know me, you assume the worst and attack out of hand. Way to jump in with both feet.
If you read past the first sentence, you would have seen that I'm not an idiot who is spouting off about nothing. Chrime 16 ran fine, it auto-updated to 17 and suddenly it needs 1GB of RAM to do the exact same thing. That's not simply a high water mark.
BTW, I WAS looking at RES, not VIRT, as stated in another comment in this thread which I'm sure you didn't read.
2GB of uncompressed images open simultaneously in a browser? I don't know what kind of porn sites you go to, but 2gb of uncompressed image data will not fit on your screen, no matter how many monitors you hook together. The only possible way you can have 2GB of uncompressed images open at a time is if you're working on high-resolution photo editing, not browsing the web.
Firefox 10.0 on Fedora 15 (i686 because I don't see a reason to run x86_64 on 2GB of RAM). That's with six tabs open (One a Google Docs spreadsheet) and Xmarks, AdBlock Plus, and Web Developer add-ons installed. Earlier today I had 32 tabs open for an hour while I caught up on some web comics.
I seriously don't understand how people can have Firefox blow up to 1GB of RAM. What add-ons and plug-ins do you have installed that it balloons up that big that fast?
On my other machine, I've had Firefox (Same version and OS) open for two weeks, with two windows (one with nine tabs open, the other with two) open for a week now and top shows RES is 409M, SHR is 30M, for a total of about 380M used.
Now if you'll excuse me, I have to go kick Chrome again. It's RES is 985M used and everything is getting laggy.
That argument is disingenuous and irrelevant. On the same hardware, Chrome 16 would run for seven days and only edge up to about 450MB RAM use. Firefox 10, after over two weeks of continuous operation is hovering in the 350MB range. There is no excuse for a web browser process to hit the GB mark, none.
As for 2GB of RAM being cheap, that's a poor excuse. When Chrome hits 1GB of RAM, it causes my entire system to begin to slow down. It affects Firefox, GNOME, even my terminal windows. The instant I kill it and restart, everything is happy again, until it creeps up there and starts thrashing the memory manager again.
One man's treasure is another man's trash. I've had the opposite experience. I HATE Chrome's updates. They've caused nothing but problems when it pushes out a major version update that silently breaks existing websites for no apparent reason (Chrome 15 to 16), or bloats up to 1GB of RAM within eight hours of launch (16 to 17). And I can't go back because their previous builds no longer show in their repositories.
You have essentially two choices: stay on 3.6 after EOL and deal with it, or upgrade.
Staying on 3.6 (Which I have to do one one machine because it's a G4 Mac and already has no support) is an option, but eventually, depending on what kind of websites you frequent, you may get pwn3d. But if you restrict yourself to known-good websites, and use extensions like AdBlock, FlashBlock, and possibly GreaseMonkey, you can probably coast along for years.
Upgrading to a new browser (Especially on Linux) is also not a terrible idea. Firefox 10 is actually pretty good about RAM use (Better than Chrome 17, for my uses), and you can set the interface to match Firefox 3.6 so you don't have to re-train yourself to the new look and feel. It's even a bit more snappy than Firefox 3.6, and it does have some nice features for web-centric users (Like pinned apps, and Firefox sync).
I understand the "I'm staying here" feeling, but unless you're willing to make some serious compromises, you're on your own.
1) There is. For the last ten years it has been euphamised as "innovation". 2) Depends on what you define as "the estate of Jack Valenti".
Do you honestly think we'd be in the same world if every Windows program down the line had to be approved by Microsoft prior to it being available to anyone? Do you think Firefox, Chrome, Safari, or Opera would exist if Microsoft could kill them in the cradle because they competed with IE?
If you're speaking literally about the estate of jack Valenti, then that's irrelevant. If you're talking about his legacy, or the "content owners" represented by the RIAA/MPAA and the other special-interest groups that wrote SOPA, then yes, it is. Hardware manufacturers lock boot loaders because the companies that commissioned the hardware from them told them to. Those same companies did so because content companies told THEM to lock down the device, lest some enterprising nerd out there figure out how to get access to said content.
So what if "most people" don't care about their code execution path. They won't take advantage of it anyway, so retaining it does nothing while removing it stops the people who DO contribute to the body of human knowledge from contributing, or at least raises the barrier to entry. The fact that "most people" don't want to work on their devices doesn't mean that those that do should be prevented from doing so.
Your argument about a Melissa or Slammer worm on all Android or IOS devices is also bogus. Melissa didn't need "root" access to do its job, and Slammer was facilitated by stupid programmer decisions by the original vendor. Neither of which will be abated by using a walled garden and in fact can be increased because of the perceived "safety" of the garden.
Yes, there will always be a small community of hackers, but said community will be smaller and more difficult to maintain as devices are increasingly locked down and lesser-skilled members decide it's not worth their time to break into their devices in order to realize whatever idea (however small) they have that could balloon into the next "innovation".
Speaking as a hobbyist here, I have done what you're asking about for the last eight years and I have very close to zero problems, however there was some ground work that had to be laid. Oh, disclaimer, I'm also a sys admin for a major hosting company (I won't tell you who) so my definition of "easy" may not match yours.
0) If you're hosting at home, make sure you have an ISP that doesn't suck (i.e. use a local ISP). I use a local ISP that has DSL/FTTC (If you happen to live in an area served by the FTTC) connections), so I pay the local ILEC for a DSL line and the ISP for the connection. If you're hosting at a VPS/colo, make sure you pick a good one that will help you out: sell you a dedicated IP AND either give you control of your reverse DNS or setup your reverse DNS for you to your specifications. This is actually a critical part. I have stayed on a 1.5Mbit DSL line for years because anything faster in my area removes my choice of ISP and that is unacceptable to me.
1) Setup your preferred e-mail infrastructure. A dedicated VM/box (I'm using a low-powered Via C7-based server. It draws 30W and handles more than just my e-mail) with whatever SMTP/IMAP server you want. I happen to use Postfix/Dovecot tied together with Procmail so I can do my own mail rules and interpose SpamAssassin in the chain to catch all the crap that comes in. You can use whatever MTA/MDA you want, but if you plan on using webmail, you should probably have an IMAP server for it to feed from.
a) Make your SMTP/IMAP servers secure. This involves setting up TLS/SSL, creating some certs (Self-signed is okay if any users of the system know to accept the "unverified" cert the first time they connect), and enabling SMTP AUTH. There are HOWTOs aplenty on how to do this.
b) Setup your filtering software, because spam handling is all on you. Personally I use SpamAssassin with whitelists/blacklists, some customized scoring for built-in rules, and a very old bayesian filter (Old meaning it has been in use for years and is very well trained, not that it has been neglected for years). I have
2) Contact your ISP or VPS/colo provider to make sure they don't filter ports and to request changes to your reverse DNS entries. I have a single IP address and I had problems with organizations like Comcast blocking me until I had my reverse DNS changed to my domain name.
3) Make sure you aren't part of the problem. I have a cron job that alerts me if I start picking up a lot of "undeliverable" bounce messages and I clean them out so I'm not annoying other mail admins by repeated attempts to deliver crap. You should also check various RBLs to make sure your IP isn't on any of their lists and if it is, contact them to get removed. I haven't had any problems with RBLs in years due to steps 0-2.
4) Happy mailing. Like I said in the beginning, I've used this setup for years and it has worked for me and several others. I routinely e-mail people on Yahoo, Gmail, Hotmail, Comcast, Cox, and other big ISPs and have not had any significant problems in years.
Notes I've tried several webmail programs, squirrelmail, roundecube, even some "groupware" and "group office" ones, they all have their pros and cons, but they all talk IMAP on their back end, and you may end up having multiple devices accessing your e-mail like I do (desktop, laptop, phone, tablet, etc).
To keep your bayesian filter updated, make sure you have a "spam" and "not spam" folder. Personally I have three, a "really spam" folder (For anything scoring 20 or higher), a "probably spam" folder (For anything marked as spam), and a "not spam" folder for anything that was mistakenly marked as spam. I almost never have to think about the "really spam" folder because I have never found anything in there that shouldn't be. I do go through the "probably spam" folder weekly looking for false-positives, and there have been a few. Those get moved to the "not spam" folder. Weekly I have a script
It's not just Gmail, Calendar, etc. It's the Market app itself also. That's the lynch pin. If you can't get the Market app on the phone, how are you going to get easy (customer-friendly) access to the rest of the things (Google-owned or otherwise) that you want? Sure geeks can side-load apps into Android devices, but non-geeks won't in any real numbers.
I used DESQview as well, and I ran a little 2-node BBS under it. I was solidly in the "love it" crowd. It ran on my 486dx33 w/4MB RAM and I was able to do up to four things at once, but typically would limit myself to 2 to save RAM.
I tried DesqView/X when it came out, but it wanted like 8MB RAM and didn't run very well, and I didn't have X Windows deployed elsewhere for it to make any real difference. I still have a DesqView/X book around somewhere.
Not that much (to version 9). RHEL doesn't do major version bumps for components before the whole OS bumps. However you will probably be able to find a community-maintained repo with PG9 packages for RHEL6 shortly (There may be one for RHEL5 now), google is your friend.
Really? Strange. I wasn't comparing ESXi to KVM, I was comparing VMware Server to ESXi. Personally, I run KVM, with KSM and VDE. I also wonder how VMware server "makes sales for ESXi" when ESXi is also free-to-download.
Yes, that's what I'm saying, and the fact that it runs as a program under an existing OS is why it isn't as fast or efficient as ESXi which is a "bare metal" hypervisor.
VMware Server and VMware ESXi are different. With ESXi, your VM server is basically a specialized appliance which runs VMs. The hypervisor isn't a regular Linux OS. It just shows a text-based message telling you to use the client or a browser to connect and manage VMs. You can ssh into the box after enabling some "maintenance mode" tweaks but even then it's not a full Linux OS, more like an embedded Linux OS since it makes heavy use of BusyBox to replace most of the command-line tools.
For enterprises, dedicating one or more physical machines to being VM servers isn't a problem, but for home use, not being able to use the VM server like a normal computer in addition to it being a VM server can be a deal-breaker.
I've been looking for something like this for a while, especially as my device herd has grown. What software would you use on the server side? I've looked into several (SOGo, eGroupware, DAVIcal, etc) but they all break in some form or other. The closest I've ever gotten is a mostly-sync (Contacts and Notes) with with eGroupware but it throws a very opaque error when I try to sync my calendar items (gets about 20 of 401 then throws an "invalid server address" error).
I'm nearly there with you, but my experience has been somewhat different.
I rolled out my own MythTV backend (First as a VM on my workstation then promoted to an MSI Wind (Atom 330, 2GB RAM, 500GB disk)) connected to a HDHomerun which is connected to a large OTA antenna mounted in my attic. Picture quality has been really good on channels that have decent video sizes and bit rates, and most of the TV transmitters are located on one mountain top, so I don't have to move the antenna to pick up different channels. I looked into cable and satellite when I was in the planning stages, but I did not want to do a D-A-D conversion to get it into Myth and the only channels my local cable monopoly provides unencrypted are the locals anyway.
My main problem is that the kind of TV I like, I can't find, even on torrent trackers. There used to be one that specialized in it (DigitalDistractions), but they went dark a few years ago and every other tracker website I've been able to poke through doesn't have it or doesn't have any shows that I don't already have. And I guess I'm not cool enough to be invited to the private trackers.
I've found a few that are almost suitable replacements, and when they air on a local HD channel, I get them in 480/720/1080, which looks acceptable to really good, and Myth brings them in usually faster than I can watch them, so I always have new content to keep me entertained.
Since the situation is so hobbled (Old Linux kernel, no LVM) about the only thing you will be able to do is learn to use hardlinks. The ext* filesystems support them but you will have to manage them yourself (cp -varl/source/*/destination/version). Yeah it's a huge hack, but unless you can actually fix the problem, it's about your only hope.
"Quad Band" means it supports 850/900/1800/1900 MHz frequencies, and all of those are GSM (voice) frequencies, not 3G (HSDPA/USM) frequencies. Unless it says 1700/2100 MHz for data access, T-Mo won't work. It's probably 850/2100MHz USM/CDMA which is what AT&T uses.
(Says the guy who owns a factory Nokia phone which only gets EDGE because of the above issue).
How do you figure? "PAE" is so easy as to be a "gimmie" outside of the Windows world. And the company I work for just finished virtualizing almost 70 servers, only 10 of which are rigged for more than 4GB of RAM.
Slashdot Mirror
True, but I was referring to a work desktop computer. I get to run desktop software (browser, office apps, terminals, e-mail, development IDE, etc) not encoding software or number-crunching. The only time it's not waiting for input is when the programs are loading, either themselves or data. About the most intensive data it gets to process (In the CPU) are encrypted e-mails or ssh sessions.
Thanks for your insightful and helpful remarks. Sarcasm aside, you don't know me, you assume the worst and attack out of hand. Way to jump in with both feet.
If you read past the first sentence, you would have seen that I'm not an idiot who is spouting off about nothing. Chrime 16 ran fine, it auto-updated to 17 and suddenly it needs 1GB of RAM to do the exact same thing. That's not simply a high water mark.
BTW, I WAS looking at RES, not VIRT, as stated in another comment in this thread which I'm sure you didn't read.
2GB of uncompressed images open simultaneously in a browser? I don't know what kind of porn sites you go to, but 2gb of uncompressed image data will not fit on your screen, no matter how many monitors you hook together. The only possible way you can have 2GB of uncompressed images open at a time is if you're working on high-resolution photo editing, not browsing the web.
Firefox 10.0 on Fedora 15 (i686 because I don't see a reason to run x86_64 on 2GB of RAM). That's with six tabs open (One a Google Docs spreadsheet) and Xmarks, AdBlock Plus, and Web Developer add-ons installed. Earlier today I had 32 tabs open for an hour while I caught up on some web comics.
I seriously don't understand how people can have Firefox blow up to 1GB of RAM. What add-ons and plug-ins do you have installed that it balloons up that big that fast?
On my other machine, I've had Firefox (Same version and OS) open for two weeks, with two windows (one with nine tabs open, the other with two) open for a week now and top shows RES is 409M, SHR is 30M, for a total of about 380M used.
Now if you'll excuse me, I have to go kick Chrome again. It's RES is 985M used and everything is getting laggy.
That argument is disingenuous and irrelevant. On the same hardware, Chrome 16 would run for seven days and only edge up to about 450MB RAM use. Firefox 10, after over two weeks of continuous operation is hovering in the 350MB range. There is no excuse for a web browser process to hit the GB mark, none.
As for 2GB of RAM being cheap, that's a poor excuse. When Chrome hits 1GB of RAM, it causes my entire system to begin to slow down. It affects Firefox, GNOME, even my terminal windows. The instant I kill it and restart, everything is happy again, until it creeps up there and starts thrashing the memory manager again.
One man's treasure is another man's trash. I've had the opposite experience. I HATE Chrome's updates. They've caused nothing but problems when it pushes out a major version update that silently breaks existing websites for no apparent reason (Chrome 15 to 16), or bloats up to 1GB of RAM within eight hours of launch (16 to 17). And I can't go back because their previous builds no longer show in their repositories.
You have essentially two choices: stay on 3.6 after EOL and deal with it, or upgrade.
Staying on 3.6 (Which I have to do one one machine because it's a G4 Mac and already has no support) is an option, but eventually, depending on what kind of websites you frequent, you may get pwn3d. But if you restrict yourself to known-good websites, and use extensions like AdBlock, FlashBlock, and possibly GreaseMonkey, you can probably coast along for years.
Upgrading to a new browser (Especially on Linux) is also not a terrible idea. Firefox 10 is actually pretty good about RAM use (Better than Chrome 17, for my uses), and you can set the interface to match Firefox 3.6 so you don't have to re-train yourself to the new look and feel. It's even a bit more snappy than Firefox 3.6, and it does have some nice features for web-centric users (Like pinned apps, and Firefox sync).
I understand the "I'm staying here" feeling, but unless you're willing to make some serious compromises, you're on your own.
1) There is. For the last ten years it has been euphamised as "innovation".
2) Depends on what you define as "the estate of Jack Valenti".
Do you honestly think we'd be in the same world if every Windows program down the line had to be approved by Microsoft prior to it being available to anyone? Do you think Firefox, Chrome, Safari, or Opera would exist if Microsoft could kill them in the cradle because they competed with IE?
If you're speaking literally about the estate of jack Valenti, then that's irrelevant. If you're talking about his legacy, or the "content owners" represented by the RIAA/MPAA and the other special-interest groups that wrote SOPA, then yes, it is. Hardware manufacturers lock boot loaders because the companies that commissioned the hardware from them told them to. Those same companies did so because content companies told THEM to lock down the device, lest some enterprising nerd out there figure out how to get access to said content.
So what if "most people" don't care about their code execution path. They won't take advantage of it anyway, so retaining it does nothing while removing it stops the people who DO contribute to the body of human knowledge from contributing, or at least raises the barrier to entry. The fact that "most people" don't want to work on their devices doesn't mean that those that do should be prevented from doing so.
Your argument about a Melissa or Slammer worm on all Android or IOS devices is also bogus. Melissa didn't need "root" access to do its job, and Slammer was facilitated by stupid programmer decisions by the original vendor. Neither of which will be abated by using a walled garden and in fact can be increased because of the perceived "safety" of the garden.
Yes, there will always be a small community of hackers, but said community will be smaller and more difficult to maintain as devices are increasingly locked down and lesser-skilled members decide it's not worth their time to break into their devices in order to realize whatever idea (however small) they have that could balloon into the next "innovation".
Speaking as a hobbyist here, I have done what you're asking about for the last eight years and I have very close to zero problems, however there was some ground work that had to be laid. Oh, disclaimer, I'm also a sys admin for a major hosting company (I won't tell you who) so my definition of "easy" may not match yours.
0) If you're hosting at home, make sure you have an ISP that doesn't suck (i.e. use a local ISP). I use a local ISP that has DSL/FTTC (If you happen to live in an area served by the FTTC) connections), so I pay the local ILEC for a DSL line and the ISP for the connection. If you're hosting at a VPS/colo, make sure you pick a good one that will help you out: sell you a dedicated IP AND either give you control of your reverse DNS or setup your reverse DNS for you to your specifications. This is actually a critical part. I have stayed on a 1.5Mbit DSL line for years because anything faster in my area removes my choice of ISP and that is unacceptable to me.
1) Setup your preferred e-mail infrastructure. A dedicated VM/box (I'm using a low-powered Via C7-based server. It draws 30W and handles more than just my e-mail) with whatever SMTP/IMAP server you want. I happen to use Postfix/Dovecot tied together with Procmail so I can do my own mail rules and interpose SpamAssassin in the chain to catch all the crap that comes in. You can use whatever MTA/MDA you want, but if you plan on using webmail, you should probably have an IMAP server for it to feed from.
a) Make your SMTP/IMAP servers secure. This involves setting up TLS/SSL, creating some certs (Self-signed is okay if any users of the system know to accept the "unverified" cert the first time they connect), and enabling SMTP AUTH. There are HOWTOs aplenty on how to do this.
b) Setup your filtering software, because spam handling is all on you. Personally I use SpamAssassin with whitelists/blacklists, some customized scoring for built-in rules, and a very old bayesian filter (Old meaning it has been in use for years and is very well trained, not that it has been neglected for years). I have
2) Contact your ISP or VPS/colo provider to make sure they don't filter ports and to request changes to your reverse DNS entries. I have a single IP address and I had problems with organizations like Comcast blocking me until I had my reverse DNS changed to my domain name.
3) Make sure you aren't part of the problem. I have a cron job that alerts me if I start picking up a lot of "undeliverable" bounce messages and I clean them out so I'm not annoying other mail admins by repeated attempts to deliver crap. You should also check various RBLs to make sure your IP isn't on any of their lists and if it is, contact them to get removed. I haven't had any problems with RBLs in years due to steps 0-2.
4) Happy mailing. Like I said in the beginning, I've used this setup for years and it has worked for me and several others. I routinely e-mail people on Yahoo, Gmail, Hotmail, Comcast, Cox, and other big ISPs and have not had any significant problems in years.
Notes
I've tried several webmail programs, squirrelmail, roundecube, even some "groupware" and "group office" ones, they all have their pros and cons, but they all talk IMAP on their back end, and you may end up having multiple devices accessing your e-mail like I do (desktop, laptop, phone, tablet, etc).
To keep your bayesian filter updated, make sure you have a "spam" and "not spam" folder. Personally I have three, a "really spam" folder (For anything scoring 20 or higher), a "probably spam" folder (For anything marked as spam), and a "not spam" folder for anything that was mistakenly marked as spam. I almost never have to think about the "really spam" folder because I have never found anything in there that shouldn't be. I do go through the "probably spam" folder weekly looking for false-positives, and there have been a few. Those get moved to the "not spam" folder. Weekly I have a script
It's not just Gmail, Calendar, etc. It's the Market app itself also. That's the lynch pin. If you can't get the Market app on the phone, how are you going to get easy (customer-friendly) access to the rest of the things (Google-owned or otherwise) that you want? Sure geeks can side-load apps into Android devices, but non-geeks won't in any real numbers.
I used DESQview as well, and I ran a little 2-node BBS under it. I was solidly in the "love it" crowd. It ran on my 486dx33 w/4MB RAM and I was able to do up to four things at once, but typically would limit myself to 2 to save RAM.
I tried DesqView/X when it came out, but it wanted like 8MB RAM and didn't run very well, and I didn't have X Windows deployed elsewhere for it to make any real difference. I still have a DesqView/X book around somewhere.
Not that much (to version 9). RHEL doesn't do major version bumps for components before the whole OS bumps. However you will probably be able to find a community-maintained repo with PG9 packages for RHEL6 shortly (There may be one for RHEL5 now), google is your friend.
If you install EPEL you'll get an additional 4600+ packages.
However RHEL/CentOS are server operating systems, so a lot of packages that make sense on desktops are omitted.
I'm curious now that you've mentioned it, have you tried VM live migration in KVM? I don't have enough hardware to do it myself.
Really? Strange. I wasn't comparing ESXi to KVM, I was comparing VMware Server to ESXi. Personally, I run KVM, with KSM and VDE. I also wonder how VMware server "makes sales for ESXi" when ESXi is also free-to-download.
Yes, that's what I'm saying, and the fact that it runs as a program under an existing OS is why it isn't as fast or efficient as ESXi which is a "bare metal" hypervisor.
VMware Server and VMware ESXi are different. With ESXi, your VM server is basically a specialized appliance which runs VMs. The hypervisor isn't a regular Linux OS. It just shows a text-based message telling you to use the client or a browser to connect and manage VMs. You can ssh into the box after enabling some "maintenance mode" tweaks but even then it's not a full Linux OS, more like an embedded Linux OS since it makes heavy use of BusyBox to replace most of the command-line tools.
For enterprises, dedicating one or more physical machines to being VM servers isn't a problem, but for home use, not being able to use the VM server like a normal computer in addition to it being a VM server can be a deal-breaker.
I've been looking for something like this for a while, especially as my device herd has grown. What software would you use on the server side? I've looked into several (SOGo, eGroupware, DAVIcal, etc) but they all break in some form or other. The closest I've ever gotten is a mostly-sync (Contacts and Notes) with with eGroupware but it throws a very opaque error when I try to sync my calendar items (gets about 20 of 401 then throws an "invalid server address" error).
What software do you use to sync? I've also got an E63 and I'm thinking of getting an E72/E73 but I want something that will sync with Evolution.
I'm nearly there with you, but my experience has been somewhat different.
I rolled out my own MythTV backend (First as a VM on my workstation then promoted to an MSI Wind (Atom 330, 2GB RAM, 500GB disk)) connected to a HDHomerun which is connected to a large OTA antenna mounted in my attic. Picture quality has been really good on channels that have decent video sizes and bit rates, and most of the TV transmitters are located on one mountain top, so I don't have to move the antenna to pick up different channels. I looked into cable and satellite when I was in the planning stages, but I did not want to do a D-A-D conversion to get it into Myth and the only channels my local cable monopoly provides unencrypted are the locals anyway.
My main problem is that the kind of TV I like, I can't find, even on torrent trackers. There used to be one that specialized in it (DigitalDistractions), but they went dark a few years ago and every other tracker website I've been able to poke through doesn't have it or doesn't have any shows that I don't already have. And I guess I'm not cool enough to be invited to the private trackers.
I've found a few that are almost suitable replacements, and when they air on a local HD channel, I get them in 480/720/1080, which looks acceptable to really good, and Myth brings them in usually faster than I can watch them, so I always have new content to keep me entertained.
[citation needed]
Seriously. I'd love an N900 without switching carriers or selling organs. Show me where I can get one.
Since the situation is so hobbled (Old Linux kernel, no LVM) about the only thing you will be able to do is learn to use hardlinks. The ext* filesystems support them but you will have to manage them yourself (cp -varl /source/* /destination/version). Yeah it's a huge hack, but unless you can actually fix the problem, it's about your only hope.
"Quad Band" means it supports 850/900/1800/1900 MHz frequencies, and all of those are GSM (voice) frequencies, not 3G (HSDPA/USM) frequencies. Unless it says 1700/2100 MHz for data access, T-Mo won't work. It's probably 850/2100MHz USM/CDMA which is what AT&T uses.
(Says the guy who owns a factory Nokia phone which only gets EDGE because of the above issue).
Can you check SMART status and issue SMART commands to individual drives? I've heard that some/many don't.
How do you figure? "PAE" is so easy as to be a "gimmie" outside of the Windows world. And the company I work for just finished virtualizing almost 70 servers, only 10 of which are rigged for more than 4GB of RAM.