At 300 players, I would highly recommend you scrap the ramdisk altogether and get an SSD. While I am a huge proponent of ramdisks, the difference between the common admin with 10 players and one who gets 150-300 is a pretty large difference--enough that an SSD is unquestionably the best solution here. Ramdisk performance is great, but when you consider holding the entire world that 300 people can explore, it's more wasteful than helpful (unless you have your world bounded pretty small).
Then, you can have nearly all your ram go to handling the high traffic Minecraft users, prevent ram-depleted crashes, and still have the performance of a high perf SSD.
Well, have you ever used /reload?
That is notorious for memory leakage, since it creates new plugin instances.
I think it should be removed, but regardless, try adding around 4gigs to your RAM disk, you typically want around 50 megabytes available per player, and 1gig available for the server, so you're near maxing your capacity :laugh.gif:
Well, we don't know how large his ramdisk CURRENTLY is, but I can't see that adding 4gb to the ramdisk (without adding 4gb of physical ram!) is going to do to help.
If the OP gets 300 users x 50mb, that's 15000--as you say-but I don't know where he'd get 4 more gigs for the ramdisk then. Assuming his ramdisk is only sized 1gb now (which is probably lowballing it), you're recommending a 5gb ramdisk, leaving 11gb for the OS and Minecraft server. Being generous and saying the OS only requires .5gb, you should also be advising he lower his max-players to about 210 (or at least some adjustment, assuming the available ram would be decreased 4gb).
That said, if the ramdisk gets too large (and java -Xmx is set), that means people are exploring and with more than 150 people doing that (theoretically), SSD is still a better choice.
(and if you can get 300 people on a server, OP, you can find a way to raise the money for it!)
No big deal, I've been backing up far more than the average person today. I go to the "Restore" page to find... The backups are from Thursday. I SSH into my server and copy the world files from Backup to Servers and start it up to find that everything that I've done today (8+ hours of work) has been lost, along with everything my friends have done. Without the backup / restore feature properly working, I will have no choice but to revert back to using the standard jar files or Bukkit, which I do NOT want to do. If this is a stupid user-error, please let me know so this won't happen again, but if this a problem, I need to know how to fix it because all the users of my servers are starting to get quite frustrated.
Thank you in advance for any advice.
So I guess the first question is...what method were you using for making your backups? Were you using the cron-tab options available from the web-ui exclusively or were you ever using any other method (e.g., ssh+cp or ssh+python scripts)?
The first concern is whether, in SSH you were doing any actions as 'root'. If you were (and you shouldn't be), many backed up files will be created/overwritten and owned by root--these files are then inaccessible for the most part to the user 'mc' which the web-ui runs on. In addition, backups made after such an action would also be problematic because all the rsync operations work as 'mc' as well.
If you can dismiss that as a possibility, the next question is in doing hourly backups, did the backups ever work as intended (every hour) with the restore points showing up in the restore tab? If so, does it seem to have arbitrarily stopped sometime thursday (such as, it was doing it hourly until X oclock and there was just nothing?)
Every hour should accrue a restore point visible in the restore panel, as well as every click on the restore.
On a side note, what do you mean this will force you to rever to the standard jar files or bukkit--the use of ramdisk or hdd should be independent from the jar used; they shouldnt interfere in any combination. Let me know about the above and we can look further into it. Also, take a trip over to the google groups forum instead which is looked after much more frequently (and have other crux users available and willing to help).
MineOS CRUX most certainly supports WPA2, which is the current configuration I have at my home network with this flimsy 10 year old laptop. I grudgingly admit it took me a while to get it up and running as well, since command line (CLI) configuration of wireless is assuredly more difficult than having windows popping up and it auto-guessing protocols, etc.
Then the whole business of turning on the device (ifconfig eth1 up) after configuring it (wpa_supplicant) and etc. DHCP is what I used, though it may be less desirable to do so for servers (unless you use static dhcp like i recommend) as well--all those on top of the necessity for wpa_passphrase. And on a GUI linux its as simple as 'type in ssid, type in pw'.
To minimize the RAM footprint, just change the INITIAL_MEMORY and MAXIMUM_MEMORY variables to something smaller than 1GB. I don't know how much memory is actually needed by the server (it probably depends on world size and the number of add ons installed if running bukkit), so you will have to play around with the values. If you want to try the options used in the link that you provided, you could change the SERVER_COMMAND variable to something like:
Note that I haven't tested this, and that as the author states you must be running the java 1.7 jdk.
I regret returning over time to be a negative nancy, but I REALLY encourage admins to not use that line. That line has many, many counter-productive, specifically detrimental arguments that are certain to interrupt stability rather than squeeze out performance.
To preface, I think it should be stated that nowhere, nobody has come up with a good set of arguments for java that even a dozen people could support cause improvements, and thats because most of the time people read what an argument does and assume they know the value better than Oracle, who sets these defaults; and the bukkit devs who know how they're programming it; and also better than those who understand regular commands like ionice and nice.
To save a lot of reading, I can just emphasize now: do not use those 'optimizations'
On this page, posts from both 'Will' and 'EvilSeph' (on all four pages) repeatedly demonstrate how many of these optimizations are not appropriate for general, blind application to a server. Specifically, EvilSeph says:
-Xnoclassgc - this flag should not be used with CraftBukkit as we make EXTENSIVE use of reflection and could lead to a large memory leak as a result of all the dynamic classes not being garbage collected when they need to be, since you are telling it not to…
If somebody with 100 'optimizations' says 'use it' and the developer of bukkit says DON'T, I hope we can agree that bukkit developers probably know better.
There's also the idea that deliberately setting XMS lower than XMX comes with its own performance costs too (explained in the linked thread).
Not to mention IONICE and NICE should never be used the same command--IONICE is already going to take precendence, making NICE useless. But the worst part is making IONICE class 1 (realtime)--application servers are NEVER supposed to be set to realtime. "realtime" sounds nice, as it feels like that's how a game should run--but that's not the proper context for running processes in realtime, which is extremely dangerous. Dangerous how? Well, when you put java as the most important process on the system, and java does the things it does like take 100% CPU, having the realtime scheduling means it will take LITERALLY 100% of the CPU... and spare no cycles for other processes lower than "-c1". When Minecraft does its notorious 100% cpu hogging, other scheduling tasks--like ones the kernel needs to do--or other processes like mysql or SSHD--these all will take the backseat both processor wise AND HDD-wise...meaning they can fail just so Minecraft can squeeze out a few more CPU cycles it doesn't need (such as in 100% CPU locks).
I have a feeling this is going to turn into another mac os vs. windows vs. linux flamewar, but I'd seriously be interested to know what's best?
I saw this quote on another thread here:
It made me think... what is the best?
I use Bukkit, not the normal Miencraft Server, so I'm curious to know what is the best for hosting a minecraft server?
Right now I use Windows Server 2003 and I'm happy with it, just curious what I'm missing out on on another OS.
thanks
It's largely accepted that there is no 'best' OS. At least, not by simply saying 'best' and not using more descriptive qualifiers, such as 'which OS requires the least amount of configuration know-how to get working?' or 'which OS has the least amount of RAM overhead?' or 'which is actually a server OS?'
That is, any novice who has used Windows desktop his or her entire life will have no problem setting up a windows server running the minecraft server executable. Is it ideal? maybe not. But for somebody with no other knowledge in hosting and administration, running that .exe off their windows server's desktop succeeds at their needs.
When you consider RAM overheads that a GUI adds, many would point toward Linux, touting Ubuntu, CentOS, or Debian as 'the OS'--of course, when one lacks the patience and aptitude to handle command-line, the 'best OS' might instead become 'the worst OS'. Just like if somebody uses a particular distro, fails at one thing, then asserts 'WORST OS EVER'.
MacOSX, in the form that probably everybody has (which is it is their main computer), isn't a server platform. Not to say it can't host, like a windows desktop could, but it certainly couldn't be the best OS on account of it simply just not being tailored for server duties. Performance per megabyte, or by pretty much any other metric, OSX probably isn't going to be the best...unless one has no idea how to use windows or linux, and it is the most simple.
These threads (or more specifically, the voting part of this thread) is unlikely to be representative of the real 'best' OS just for these biases mentioned above, but all that aside, most people will answer 'what is the best OS' with 'what OS did I have just the right amount of patience to stick with?'
I think there are a few points worth summarizing/expounding on; unfortunately, it won't bring back a lost server or an unreasonable not-being-allowed-to-host.
1) Even assuming there are simple tools to just 'hack a server' and compromise everything on the server computer itself, the probability of being a target is so appallingly low. There are literally thousands of servers out there hosting Minecraft all of which have so much more appeal to be hacked (that is, servers that have information/data WORTH stealing). People would not, quite frankly, waste their time going after friends-only servers. To me, that is the single-most compelling argument that hosting from home is not dangerous.
2) A server that isn't running can't possibly be dangerous. To have to delete the files even after being told not to host is not only unreasonable, but it's also useless. There's no technology or technique in the world that can exploit a server if the server isn't accessible (i.e., not running). Once you shut down the server ("stopped" it), it's about as exploitable as Minesweeper.
3) Unfortunately, you don't have much hosting potential anymore being that you were banned from doing so, but another approach that could have ensured the safety of your system would have been to run it in a VM. Whether Windows or Linux (preferably so), if someone could easily hack Minecraft to gain control of a system, gaining control of a VM dedicated to Minecraft means their efforts would be mostly fruitless.
That said, sorry about losing the server and your work.
Hello, I just wanted to say that I just recently (about 4-5 days ago) installed your MineOS CRUX on a laptop I had lying around. It is not that old of a laptop but the graphics card on it completely fried and I cannot run any graphical OS on it at all.. that was when I found your OS and decided that I would give it a shot running my minecraft server. Well, even with a fried graphics card (which makes it completely unusable for pretty much everything) it runs the server beautifully! I just have a couple quick questions that may/maynot be relevant at all..
During the install you create two different partitions, I have a 250GB HD and I want as much space as possible for the backups/worlds. I wasn't exactly sure which partition to make as the biggest one, so (as in your tutorial video) I created the first partition with 245GB and the second with 5GB thinking the first partition you create was the main one. Am I correct in assuming this? or should I readjust my partitions to make the second one larger?
Also, I was wondering of you could provide an option in the web-ui for downloading the latest stable development build of craftbukkit so that we could create a separate server to test it on just to double check that it will work with our current server. Just a suggestion :smile.gif:
And again, thank you for your awesomeness!
Glad to see MineOS is working out for you! As far as the biggest one, the appropriate partition to have that (is the sda1 -- the first partition -- in the setup). This is where the OS is set up and the world files are stored by default. All other things downloaded or used will end up on this partition. Your 5gb will be the swap partition.
Note, you still can run two or more builds (recommended and dev bukkits) in parallel (and simultaneously if you like), but the UI will only update one: the recommended one. Tthis way I, as the MineOS developer, never introduced new, plugin-breaking updates to their system via a system update--that users had to be specifically aware of deciding to use dev versions. As noted in the wiki, if you're ready to use the 1.0.0 for full from the UI, there are instructions to doing so.
If you have any other questions, feel free to drop into the google groups page where most of the people go for MineOS CRUX community support https://groups.google.com/forum/#!forum/mineos
Anyone viewing this is welcome to take my idea and make this plugin but you must note in your plugin's post that it was my idea.
This part is crucial. We definitely need to give credit where credit is due. And the inspiration of the 'auto-ban plugin for using a set of annoying spam words' is what I know I would like MY legacy to be.
@ccundiff12, MineOS was pretty cool for about 2 weeks. Then for some reason it will not boot the server. So I would not recommend it to anymore at this point. The support for it is very limited.
Minecraft server files check:
invalid number of arguments
Based on that error, you were messing with the mineos.config file. Either you somehow deleted it or you messed up one of five easily identifiable lines found within. If it stopped working after two weeks 'without explanation', it was your user-error.
If you believe there is limited support, you didn't try very hard, as I have never seen anybody report this issue, I've never been contacted about it on my forums, google groups or personal email, and quite frankly, it took all of ten seconds to code-trace exactly where this issue could have arisen. There may be fewer people to 'support' it, but unlike the hundreds of Linux experts on these forums, my answers are always diligently researched and comprehensive.
As that file path is likely to change... can another arguement be added there? say 'bukkitdev_jar =' and 'bukkitdev_jarloc =' As i have been manually updating dev builds, the webUI does still update the recommended build which will eventually work again. I can choose either jar from the web interface.
It's unlikely the old URLs will work in the future, since the old URL reflects 0.0.1, rather than the 1.0.0 which is bound to come up eventually. I'll be updating the ISO soon to reflect the recommended 1.0.0 build, but admins should expect to have to make this single change (assuming bukkit does, in fact, rename their final build).
As far as an additional argument, not likely I'll be able to add that. There's a number of places in the code that will have to change to reflect that sort of addition (more than just the config), but the benefits are meager at best.
In other words, I'm okay with admins having the necessity to know how to and be willing to download parallel dev versions and select it in the web-ui dropdown; I think come 1.0.0 bukkit, dev version updating is not going to be nearly as appealing as it was up until now, with it catching up to the final release.
Really? Awesome, going to do that then. I kinda like working with linux XD
Out of curiosity, why bother with FTP at all anyway? SFTP is less restricted in every way, installed by default, more secure, and uses the same applications (winscp/filezilla/etc)
What does FTP offer that's worth the added effort?
Three admins = $60,000USD per year if I use cheap overseas labor, and about $130,000USD if I use USA based labor.
What??!
Especially if you considered full-time admins to look after your server in a professional-sense, there is no way this model will work.
It's not that people aren't interested in mega-servers, it's that mega-servers are appealing to people only for the sake of a high population count. Near to none of the aspects of Minecraft would seem to facilitate this kind of design as a genuine business model. If I may speculate:
1) First off, there is no network of systems that could ever support this. Using one server, yogscast record breaking has shown us 899 players. This was an unplayable state. Other record breaker vids show over 2800; this was an unplayable state.
Now it may feel more plausible when you think: "well, I'll have so many more computers and so much more internet bandwidth available!" but that doesn't address the real issue: there is just too much traffic from a minecraft server.
Assuming Minecraft was already multithreaded, we'd get far better CPU throughput. But even if CPU was not the bottleneck, 100,000+ players is not practical. As a player visits a new area, he gets all the chunks surrounding him at radius R. The players around him, too, have this effect. But imagine the effect of a single block being destroyed or placed:
1) client sends request to server to destroy block
2) server validates request destroys block
3) server sends update to client / nearby people
4) server sends update to all clustered computers
It's #4 that would practically defeat the clustering system. And that's assuming the ACTUAL model shouldn't be this:
1) client sends request to server to destroy block 2a) server broadcasts request to all clustered computers 2b) server waits for validation from all clustered computers
3) server sends update to client / nearby people
4) server sends update to all clustered computers
#2 is something a multi-threaded server would do (replace computers with cores), but with the access to the information physically close (just a bus trip down), rather than physically elsewhere where latency is required. Say one server doesn't respond? How long should the server wait till it gives up on its response? Well, whatever value that is has to accommodate the latency and add it right to the amount your client will wait for a single block to be placed or destroyed. This isn't the kind of experience any player wants.
...That and the fact with 100,000+ players the sense of community is ZERO. The game will be nothing but chatlog spam, griefing will be practically impossible to avoid (without plugins), and WITH plugins, users will end up feeling like the game takes hours of travelling to find real estate (because people won't let you touch their stuff).
and where would people find resources? Well, it'd have to be unclaimed area, so I suspect there's probably a lot of running you'd do to find that....and then back to your own real estate. Need I reiterate that we're talking 100,000+ players, which compares greatly to the average large server....80-100?
In other words, nearly every aspect of Minecraft cannot support such a mega-server. (And nobody will pay for a server that gives less community and more wasted time than a non-mega-server). And accumulating 130,000USD to look after it or to oursource cheap labor...well, I should hope that already illustrates the implausibility of the idea.
[But that's just my 2c, I have never hosted a server before]
First... I'm assuming your MineOS server resides on your local network and it has been STOPPED and backed up.
Get WinSCP - http://www.winscp.net
Drag the new file from left to right. There will be two craftbukkit-x.0.x-SNAPSHOT.jar files. this is OK.
.....
now... i'm not sure how it'll keep downloading this file in the future as it updates with this filename... but i'll try and remember to comment as it becomes a problem for me.
BuildCraft Server Installer for MineOS Crux
To run the script:
1. Use wget to download it from GitHub (Wget-friendly link)
2. Place it anywhere
3. Run as root(./BuildCraftInstaller.sh)
Will it also execute properly as 'mc' rather than root? It is safer and more consistent that way.
Hey guys, just bought a server. How do I get latest craftbukkit.jar working? I've dragged it in via ftp, and now there is craftbukkit.jar and minecraft_server.jar
Any help? Much appreciated.
Buying a server is a pretty big step without having the rudimentary knowledge of starting a jar file...
0
Then, you can have nearly all your ram go to handling the high traffic Minecraft users, prevent ram-depleted crashes, and still have the performance of a high perf SSD.
Well, we don't know how large his ramdisk CURRENTLY is, but I can't see that adding 4gb to the ramdisk (without adding 4gb of physical ram!) is going to do to help.
If the OP gets 300 users x 50mb, that's 15000--as you say-but I don't know where he'd get 4 more gigs for the ramdisk then. Assuming his ramdisk is only sized 1gb now (which is probably lowballing it), you're recommending a 5gb ramdisk, leaving 11gb for the OS and Minecraft server. Being generous and saying the OS only requires .5gb, you should also be advising he lower his max-players to about 210 (or at least some adjustment, assuming the available ram would be decreased 4gb).
That said, if the ramdisk gets too large (and java -Xmx is set), that means people are exploring and with more than 150 people doing that (theoretically), SSD is still a better choice.
(and if you can get 300 people on a server, OP, you can find a way to raise the money for it!)
0
So I guess the first question is...what method were you using for making your backups? Were you using the cron-tab options available from the web-ui exclusively or were you ever using any other method (e.g., ssh+cp or ssh+python scripts)?
The first concern is whether, in SSH you were doing any actions as 'root'. If you were (and you shouldn't be), many backed up files will be created/overwritten and owned by root--these files are then inaccessible for the most part to the user 'mc' which the web-ui runs on. In addition, backups made after such an action would also be problematic because all the rsync operations work as 'mc' as well.
If you can dismiss that as a possibility, the next question is in doing hourly backups, did the backups ever work as intended (every hour) with the restore points showing up in the restore tab? If so, does it seem to have arbitrarily stopped sometime thursday (such as, it was doing it hourly until X oclock and there was just nothing?)
Every hour should accrue a restore point visible in the restore panel, as well as every click on the restore.
On a side note, what do you mean this will force you to rever to the standard jar files or bukkit--the use of ramdisk or hdd should be independent from the jar used; they shouldnt interfere in any combination. Let me know about the above and we can look further into it. Also, take a trip over to the google groups forum instead which is looked after much more frequently (and have other crux users available and willing to help).
https://groups.google.com/forum/#!forum/mineos
0
Then the whole business of turning on the device (ifconfig eth1 up) after configuring it (wpa_supplicant) and etc. DHCP is what I used, though it may be less desirable to do so for servers (unless you use static dhcp like i recommend) as well--all those on top of the necessity for wpa_passphrase. And on a GUI linux its as simple as 'type in ssid, type in pw'.
0
I regret returning over time to be a negative nancy, but I REALLY encourage admins to not use that line. That line has many, many counter-productive, specifically detrimental arguments that are certain to interrupt stability rather than squeeze out performance.
To preface, I think it should be stated that nowhere, nobody has come up with a good set of arguments for java that even a dozen people could support cause improvements, and thats because most of the time people read what an argument does and assume they know the value better than Oracle, who sets these defaults; and the bukkit devs who know how they're programming it; and also better than those who understand regular commands like ionice and nice.
To save a lot of reading, I can just emphasize now: do not use those 'optimizations'
If you are one who needs evidence, I point you toward http://forums.bukkit.org/threads/optimize-java-for-minecraft-in-a-extreme-way-p-need-feedback-thx-i-am-back.25994/page-3#post-493162
On this page, posts from both 'Will' and 'EvilSeph' (on all four pages) repeatedly demonstrate how many of these optimizations are not appropriate for general, blind application to a server. Specifically, EvilSeph says:
If somebody with 100 'optimizations' says 'use it' and the developer of bukkit says DON'T, I hope we can agree that bukkit developers probably know better.
There's also the idea that deliberately setting XMS lower than XMX comes with its own performance costs too (explained in the linked thread).
Not to mention IONICE and NICE should never be used the same command--IONICE is already going to take precendence, making NICE useless. But the worst part is making IONICE class 1 (realtime)--application servers are NEVER supposed to be set to realtime. "realtime" sounds nice, as it feels like that's how a game should run--but that's not the proper context for running processes in realtime, which is extremely dangerous. Dangerous how? Well, when you put java as the most important process on the system, and java does the things it does like take 100% CPU, having the realtime scheduling means it will take LITERALLY 100% of the CPU... and spare no cycles for other processes lower than "-c1". When Minecraft does its notorious 100% cpu hogging, other scheduling tasks--like ones the kernel needs to do--or other processes like mysql or SSHD--these all will take the backseat both processor wise AND HDD-wise...meaning they can fail just so Minecraft can squeeze out a few more CPU cycles it doesn't need (such as in 100% CPU locks).
(TLDR: keep to the simple optimizations).
1
It's largely accepted that there is no 'best' OS. At least, not by simply saying 'best' and not using more descriptive qualifiers, such as 'which OS requires the least amount of configuration know-how to get working?' or 'which OS has the least amount of RAM overhead?' or 'which is actually a server OS?'
That is, any novice who has used Windows desktop his or her entire life will have no problem setting up a windows server running the minecraft server executable. Is it ideal? maybe not. But for somebody with no other knowledge in hosting and administration, running that .exe off their windows server's desktop succeeds at their needs.
When you consider RAM overheads that a GUI adds, many would point toward Linux, touting Ubuntu, CentOS, or Debian as 'the OS'--of course, when one lacks the patience and aptitude to handle command-line, the 'best OS' might instead become 'the worst OS'. Just like if somebody uses a particular distro, fails at one thing, then asserts 'WORST OS EVER'.
MacOSX, in the form that probably everybody has (which is it is their main computer), isn't a server platform. Not to say it can't host, like a windows desktop could, but it certainly couldn't be the best OS on account of it simply just not being tailored for server duties. Performance per megabyte, or by pretty much any other metric, OSX probably isn't going to be the best...unless one has no idea how to use windows or linux, and it is the most simple.
These threads (or more specifically, the voting part of this thread) is unlikely to be representative of the real 'best' OS just for these biases mentioned above, but all that aside, most people will answer 'what is the best OS' with 'what OS did I have just the right amount of patience to stick with?'
Seconded only if you include Scientology.
0
1) Even assuming there are simple tools to just 'hack a server' and compromise everything on the server computer itself, the probability of being a target is so appallingly low. There are literally thousands of servers out there hosting Minecraft all of which have so much more appeal to be hacked (that is, servers that have information/data WORTH stealing). People would not, quite frankly, waste their time going after friends-only servers. To me, that is the single-most compelling argument that hosting from home is not dangerous.
2) A server that isn't running can't possibly be dangerous. To have to delete the files even after being told not to host is not only unreasonable, but it's also useless. There's no technology or technique in the world that can exploit a server if the server isn't accessible (i.e., not running). Once you shut down the server ("stopped" it), it's about as exploitable as Minesweeper.
3) Unfortunately, you don't have much hosting potential anymore being that you were banned from doing so, but another approach that could have ensured the safety of your system would have been to run it in a VM. Whether Windows or Linux (preferably so), if someone could easily hack Minecraft to gain control of a system, gaining control of a VM dedicated to Minecraft means their efforts would be mostly fruitless.
That said, sorry about losing the server and your work.
0
Glad to see MineOS is working out for you! As far as the biggest one, the appropriate partition to have that (is the sda1 -- the first partition -- in the setup). This is where the OS is set up and the world files are stored by default. All other things downloaded or used will end up on this partition. Your 5gb will be the swap partition.
I do not intend to provide any new options in the web ui for dev craftbukkit (since this era will soon be over, going into final 1.0 bukkit), but the instructions to use the dev 1.0 release are here: http://minecraft.codeemo.com/buildmineos/index.php/Bukkit#Auto-updating_a_different_bukkit_jar
Note, you still can run two or more builds (recommended and dev bukkits) in parallel (and simultaneously if you like), but the UI will only update one: the recommended one. Tthis way I, as the MineOS developer, never introduced new, plugin-breaking updates to their system via a system update--that users had to be specifically aware of deciding to use dev versions. As noted in the wiki, if you're ready to use the 1.0.0 for full from the UI, there are instructions to doing so.
If you have any other questions, feel free to drop into the google groups page where most of the people go for MineOS CRUX community support https://groups.google.com/forum/#!forum/mineos
0
This part is crucial. We definitely need to give credit where credit is due. And the inspiration of the 'auto-ban plugin for using a set of annoying spam words' is what I know I would like MY legacy to be.
0
Based on that error, you were messing with the mineos.config file. Either you somehow deleted it or you messed up one of five easily identifiable lines found within. If it stopped working after two weeks 'without explanation', it was your user-error.
If you believe there is limited support, you didn't try very hard, as I have never seen anybody report this issue, I've never been contacted about it on my forums, google groups or personal email, and quite frankly, it took all of ten seconds to code-trace exactly where this issue could have arisen. There may be fewer people to 'support' it, but unlike the hundreds of Linux experts on these forums, my answers are always diligently researched and comprehensive.
0
It's unlikely the old URLs will work in the future, since the old URL reflects 0.0.1, rather than the 1.0.0 which is bound to come up eventually. I'll be updating the ISO soon to reflect the recommended 1.0.0 build, but admins should expect to have to make this single change (assuming bukkit does, in fact, rename their final build).
As far as an additional argument, not likely I'll be able to add that. There's a number of places in the code that will have to change to reflect that sort of addition (more than just the config), but the benefits are meager at best.
In other words, I'm okay with admins having the necessity to know how to and be willing to download parallel dev versions and select it in the web-ui dropdown; I think come 1.0.0 bukkit, dev version updating is not going to be nearly as appealing as it was up until now, with it catching up to the final release.
0
Out of curiosity, why bother with FTP at all anyway? SFTP is less restricted in every way, installed by default, more secure, and uses the same applications (winscp/filezilla/etc)
What does FTP offer that's worth the added effort?
0
What??!
Especially if you considered full-time admins to look after your server in a professional-sense, there is no way this model will work.
It's not that people aren't interested in mega-servers, it's that mega-servers are appealing to people only for the sake of a high population count. Near to none of the aspects of Minecraft would seem to facilitate this kind of design as a genuine business model. If I may speculate:
1) First off, there is no network of systems that could ever support this. Using one server, yogscast record breaking has shown us 899 players. This was an unplayable state. Other record breaker vids show over 2800; this was an unplayable state.
Now it may feel more plausible when you think: "well, I'll have so many more computers and so much more internet bandwidth available!" but that doesn't address the real issue: there is just too much traffic from a minecraft server.
Assuming Minecraft was already multithreaded, we'd get far better CPU throughput. But even if CPU was not the bottleneck, 100,000+ players is not practical. As a player visits a new area, he gets all the chunks surrounding him at radius R. The players around him, too, have this effect. But imagine the effect of a single block being destroyed or placed:
1) client sends request to server to destroy block
2) server validates request destroys block
3) server sends update to client / nearby people
4) server sends update to all clustered computers
It's #4 that would practically defeat the clustering system. And that's assuming the ACTUAL model shouldn't be this:
1) client sends request to server to destroy block
2a) server broadcasts request to all clustered computers
2b) server waits for validation from all clustered computers
3) server sends update to client / nearby people
4) server sends update to all clustered computers
#2 is something a multi-threaded server would do (replace computers with cores), but with the access to the information physically close (just a bus trip down), rather than physically elsewhere where latency is required. Say one server doesn't respond? How long should the server wait till it gives up on its response? Well, whatever value that is has to accommodate the latency and add it right to the amount your client will wait for a single block to be placed or destroyed. This isn't the kind of experience any player wants.
...That and the fact with 100,000+ players the sense of community is ZERO. The game will be nothing but chatlog spam, griefing will be practically impossible to avoid (without plugins), and WITH plugins, users will end up feeling like the game takes hours of travelling to find real estate (because people won't let you touch their stuff).
and where would people find resources? Well, it'd have to be unclaimed area, so I suspect there's probably a lot of running you'd do to find that....and then back to your own real estate. Need I reiterate that we're talking 100,000+ players, which compares greatly to the average large server....80-100?
In other words, nearly every aspect of Minecraft cannot support such a mega-server. (And nobody will pay for a server that gives less community and more wasted time than a non-mega-server). And accumulating 130,000USD to look after it or to oursource cheap labor...well, I should hope that already illustrates the implausibility of the idea.
[But that's just my 2c, I have never hosted a server before]
0
How to move files onto your server: http://minecraft.codeemo.com/buildmineos/index.php/SFTP
And specifically, edit the file /usr/games/minecraft/mineos.config
to edit the lines to reflect the new version.
Incidentally, just by changing the mineos.config will automatically make that new file auto-download on update, so the SFTP steps arent necessary.
0
Will it also execute properly as 'mc' rather than root? It is safer and more consistent that way.
0
Buying a server is a pretty big step without having the rudimentary knowledge of starting a jar file...
http://wiki.bukkit.org/Setting_up_a_server#Linux