Communications (rahul.net)
Navigation
Dedicated Servers | Rate Chart | FAQ | Demo 1 | Demo 2 | Getting Started | Activity Links | Site Design
On This Page
Top | Index | General | Linux Choices | Installation | Mandrake Linux | After Installation | Updates & Upgrades | Hardware | Bandwidth | Bottom
Q. How reliable is this FAQ document?
This document is part of a web site and should not be read in isolation. All information is subject to change without notice. Not everything described here applies equally to every possible combination of hardware and software. All products and services are provided on an “as is” basis; “with all faults” and “as available.” THERE ARE NO WARRANTIES, EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THOSE OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. We are not responsible for the content of external web sites.
Q. What exactly is a TQMbox server?
TQMbox is a service mark of A2I COMMUNICATIONS.
Our TQMbox servers are unique in that they are more manageable than dedicated servers that you will find elsewhere. They give you full control, including the ability to install the operating system, partition disks, enter rescue mode, and, with enough skill, recover a lost boot block or partition table.
Q. Who should get a TQMbox server?
You should! But only if most of the following apply to you.
Q. Why would I want to install the operating system myself?
Our operating system install/reinstall fee for a default install is low (see Miscellaneous Fees), so you may find it affordable and convenient to ask us to install the operating system for you. This has two disadvantages for you, at least the first time around.
If you don't need any free technical support, and do not wish to become a more knowledgeable user, then we can install the operating system for you, for our low operating system install/reinstall fee. This is a good choice for customers who are looking for a professionally-managed server and responsive technical support, and are willing to pay for it. We can take over all server administration and install all needed software for you at our consulting rate.
If you install the operating system yourself, you have several choices. Each choice takes only a few minutes of your time at the keyboard (not counting the install itself, which continues on its own so you don't have to sit and wait for it). So you can try installing the operating system a few times, choosing different install options each time, and you may end up with a machine better configured to meet your needs. You can experiment with various disk partitioning schemes or (on a dual-disk server) try RAID versus no RAID. You can also explore the rescue mode—and the best time to do so is now, at the beginning of deployment of your server, while you have nothing to lose by just trying the various commands in rescue mode. This will give you more confidence in your ability to deal with operating system problems, should they ever occur in the future.
These are not virtual servers. Each is a real, independent machine, dedicated for your use only, for the duration of your service. You do not share the machine with any other customer. You get full root access and full use of the machine resources.
Software RAID provides better reliability than not using RAID. Each filesystem that is part of a RAID-1 pair is duplicated on two disks. If the operating system detects that one of the copies of the filesystem is being affected by faulty disk input/output, it will stop using that copy, but continue to function normally using only the other copy.
In some of the type R operating systems, the install procedure does not properly install the boot loader grub on the second disk of a RAID disk pair. Here are some web pages that discuss this problem and propose solutions:
Our uninterruptible and interactive install options on type R systems will automatically install grub on the second disk too, and it will also add an item to the grub boot menu that will let you boot the machine off the kernel on the second disk when needed.
Software RAID will allow uninterrupted operation in most cases where a disk failure on a non-RAID machine would lead to an unusable machine; but it's not a panacea. If the boot sectors are corrupted enough to prevent the machine from booting, you may have to go into rescue mode and attempt recovery from there.
Also note that software RAID can be more efficient, or less efficient, than no software RAID, depending on the situation. In a write-intensive environment there may be noticeable performance overhead, because the server has to work harder to write all data, in duplicate, to two disks. In a read-intensive environment software RAID may be more efficient than no software RAID, because the operating system will do some reads from one disk and some from the other, so each disk has to do fewer read operations per second. In typical web-hosting contexts, where most hits cause files to be read, software RAID performance is likely to be quite good. The best way to be sure is to do some tests with your own specific application.
Since we make it so easy to reinstall the operating system, you can do some initial tests on your newly-deployed TQMbox server with and without software RAID, and decide for yourself which you prefer for your application. Then, once you have made the decision, you can install the operating system configured appropriately for production use of the server.
Note: Our observations indicate that the installer for Mandrake Linux partitions disks erratically when software RAID is in use. See “Why aren't mirrored RAID pairs identical in size?”.
Q. Why does my web browser give me a certificate warning?
Probably because you are accessing an SSL-enabled URL (beginning with “https”), and your web browser does not know whether or not you trust the server to which you are connecting. You can permanently tell your web browser to trust our SSL-enabled web servers by installing our SSL root certificate (see Activity Links). Or you can simply skip past the warning. Below are some web sites (not connected with us) that discuss the use of SSL root certificates.
Q. Which Linux distributions do you support?
Strictly speaking, we support none of them (except for paid support). You will provide your own software support. But we do provide you with a mechanism for remotely accessing the install, update, and rescue functionality built into the Linux distribution that you will be using. We also provide a local update server that will mirror the official production quality updates that have been issued. Our operating system installation menus also let you optionally automatically set up RAID-1, i.e., mirrored filesystems on a dual-disk server.
We provide these features as shown in the table below.
| Linux Distribution | OS Install | Local Update Server | OS Rescue Mode | Automatic RAID | OS Type |
|---|---|---|---|---|---|
| RedHat 9 | Yes | Yes | Yes | Yes | R |
| Fedora Core 2 | Yes | Yes | Yes | Yes | R |
| Fedora Core 3 | Yes | Yes | Yes | Yes | R |
| Centos 3.3 | Yes | Yes | Yes | Yes | R |
| Mandrake 10 | Yes | No | Yes | Yes | M |
| Mandrake 10.1 | Yes | No | Yes | Yes | M |
| PLD Rescue CD | No | No | Yes | No | D |
The operating systems (OSs) listed as type R (which use RedHat conventions) have many similarities among themselves. The operating systems listed as type M (Mandrake conventions) and type D (Debian conventions) are different from type R and from each other in many respects. The operating system described above as PLD Rescue CD is used only for rescue operations.
The above table presents a simplified picture. Not all hardware that we supply is compatible with all software mentioned above. Our rate chart will tell you which hardware configuration has been successfully tested with which software.
Q. Can I install an operating system not on your menus?
Generally, no. However, there are reports on the Internet of people installing one operating system and then installing a second one by bootstrapping off the first one. Below are some web pages that we found. We have not tested the methods described therein.
Q. What if there are software problems?
Here are some of your choices. You are advised to practise using choices 1–7 well in advance of when you actually need them.
Q. Can I install two Linuxes on the same machine?
We were able to install two different Linuxes on a dual-disk machine using the procedure below. Try it and see how well it works for you. This procedure has been only briefly tested and is experimental. Inquire from us about having us install two Linuxes for you for a fee.
Q. Why don't I see the Initial Menu when I reboot my server?
On some hardware platforms, a normal reboot will not show you our menu. To see the Initial Menu, reboot the server while at the serial console. When you see the grub boot screen, hit the “c” key on your keyboard. You will then see a command-line prompt from grub. At this prompt type 'reboot' and hit the ENTER key. The machine will reboot to our Initial Menu.
Another less-preferred method is to log into the server as root and use the 'halt' command to halt it. Watch the serial console of the server and wait for the messages to show that the machine is halting. Wait about 15 seconds more, and then go to the Server Configuration web page and power-cycle your server. When it boots, you will see the Initial Menu. This method is less preferred because power-cycling a machine stresses the hardware and should be done only infrequently.
Q. Why is the installation password so easy to guess?
Some of the items on our operating system installation menus will ask you for a password. This password is always the word “risky” and is not really intended to be a password and provides no security. The purpose of requiring you to type this is solely to decrease the likelihood of your proceeding to the next step accidentally or absent-mindedly and causing data loss on your server. The menu items that require you to type “risky” are the ones that in our opinion are more likely than others to damage existing data on your server's disks. It is in fact possible to bypass the password prompt, by using grub's "e" command. Also, note that menu items that do not require you to type “risky” may also, if not used with sufficient care, cause data loss on your server.
Q. How can I prevent my session timing out during operating system installation?
During operating system installation, there is screen output periodically, so ordinarily your session should not time out. A connection-sharing gateway at your end, or your terminal program, may time out idle connections quite quickly. The console server at our end will also time out after 30 to 90 minutes of inactivity. You may need to type something every few minutes to keep the session active.
After the operating system installation completes, the system gives you a final static screen at which you will need to hit the ENTER key. If the machine stays at this screen for a long time, your session might time out, since no ongoing screen output is being generated. The simplest work-around for this is to not leave your terminal unattended for too long.
The well-known “putty” program has an option on its Connection screen to periodically send a null packet to keep the connection alive. A good value to use in the entry box for this is 300 seconds. Then your session will never time out.
Another option is to buy from us an Advanced shell account ($30/month plus sign-up fee). This type of account lets you use the “screen” program to create multiple virtual windows. Now you can connect to the serial console of your server from one of these virtual windows. Even if your connection across the Internet is lost, the “screen” program continues to maintain your virtual windows. Simply reconnect to our shell server and restore the “screen” session. You will still be subject to the 30-to-90–minute time-out of our console server, but any time-out at your end will not abort your session. If you lose the connection, simply reconnect and your previous serial console session will be intact. The Advanced account also includes our NW56 nationwide dial-in service.
Q. Why do you use dhcp to assign fixed IP addresses?
Our dhcp servers know your machine's static IP address, and will always assign it that same static IP address. So using dhcp, in this case, is just another way of specifying a fixed IP address for your server, and saves you the trouble of having to explicitly enter the server IP address, gateway, and name servers. This makes our operating system install methods truly automatic.
By using dhcp, we also make the rescue modes on the utility menu use the correct network information, without you having to re-enter it.
You may enter the network parameters explicitly while doing an interactive or manual install, if you wish. This makes your server independent of our dhcp servers when it boots. After the installation has completed, you can still reconfigure the server to use a static IP address, as described below. On the type R operating systems, you should be able to use the X Windows program “redhat-config-network” or “system-config-network” to do this. On the type M operating systems, webmin should already be active, so point your web browser to https://xxxx:10000/ where xxxx is the host name or IP address of your server, log in as root, and use the menus to configure the network.
Q. Which Ethernet interface should I use—eth0 or eth1?
The Ethernet interface that your server uses for its connectivity is eth0. If your machine has two Ethernet interfaces, the second interface eth1 is not normally used.
Recommended settings for type R operating systems are:
Network Device: eth0
[*] Use bootp/dhcp
[*] Activate on boot
and
Network Device: eth1
[*] Use bootp/dhcp
[ ] Activate on boot
The above settings will cause only eth0 to be activated when the machine boots, while eth1 will remain inactive. Also please see the previous question about using dhcp. If you decide to enter explicit IP information, do so only for eth0. Leave eth1 at the recommended settings shown above.
The install screens will be different for type M operating systems but you should select the closest possible equivalents; i.e., both should be configured as using dhcp, but only eth0 should be activated.
Q. What firewall settings should I use?
We provide default settings that will allow incoming ping, ssh, telnet, http, smtp, and ftp, and block just about everything else. If this is acceptable to you, and you are doing an interactive install, you can simply select “OK” at install time. These choices may not be suitable for all users, so you can use the install-time screens to customize them f you wish.
If you make the wrong choices, you may find that you cannot access the machine across the network.. If this happens, you will need to log in as root via the serial console and then open up the firewall. As an emergency fix, after logging on on the serial console port, you can simply deactivate the firewall:
# /sbin/service iptables stop
This deactivates the firewall until the next reboot. With the firewall off, your server is more exposed to attacks from the Internet, so you should fix any misconfiguration soon and then either reboot the machine or explicitly re-activate the firewall:
# /sbin/service iptables start
Q. Can I add languages other than English to my server during operating system install?
You should be probably be able to, but we have not tested this and can provide no technical support for this. If you do add any other languages during install time, please check to make sure that your “ssh” client can display the utf-8 character set, otherwise you may see corrupted screen output during the install procedure.
Q. Can I specify a different timezone during operating system install?
The server is physically in the Pacific time zone (“America/Los_Angeles” on the install screen), and this is chosen by us as the default for uninterruptible and interactive installs. During interactive or manual install, you may specify any time zone of your choice—it affects only how time is displayed, not how the kernel keeps track of it. In fact, the kernel has no knowledge of the time zone that you have set, and always uses universal coordinated time (UTC). Only time/date library functions called from user programs know about the time zone. So when you type “date”, for example, the “date” program gets the UTC time from the kernel, and then calls a library function to convert it to your preferred time zone, only for displaying it to you.
Q. Which package groups should I install?
Our supplied defaults will include most of the available packages. In type R operating systems, this is equivalent to selecting “Everything”. In type M operating systems, our default includes as many packages as the installer can handle.
Our default is a reasonably safe choice that should be satisfactory to most users. You should install less than our default only if you have good reasons for doing so. For example: if you are doing repeated operating system installs, just for practice, it may be useful to install only a limited set of packages, to speed up the install time.
To install a limited set of packages, you have to do an interactive or manual install. The uninterruptible install always uses our default, which errs on the side of installing too much rather than too little.
Q. Why does installation abort with an error?
A hardware problem can cause installation errors. But more likely there is a simpler cause. Check for these possibilities.
Q. When it asks me about monitor settings, or X configuration, what should I do?
Monitor: If asked, leave settings at the default of “unprobed monitor” and then, if asked, select “proceed”. X configuration: If asked, specify “Skip X Configuration”. If you do not see exactly these choices on the menu, look for something reasonably equivalent.
Q. Can I customize the operating system installation even more than your menus allow?
Yes, if you are technically knowledgeable enough to do so. Look carefully at the “kernel” line that appears on the screen just after you select one of the install options from our menu.
For the type R operating systems, it will be similar to this:
kernel /rh9-vmlinuz ... \ method=http://192.160.13.6/rh9os ... \ ks=http://192.160.13.6/<some pathname>
The backslashes above are line continuation characters, indicating that the three physical lines above are really the same line.
The first URL on the line (“method=...”) is the location of the operating system distribution from where files will be downloaded and installed. The second URL (“ks=...”) is the location of a Kickstart file in the format specified by RedHat. You may replace either, or both, with URLs pointing to your own http or ftp sites. To edit the line, instead of hitting the ENTER key while at the grub menu, hit the “e” key instead. Then you will see grub's edit menu, from which you may select lines to edit before you boot the system. The procedures to do this are outside the scope of our free support, but they are well-documented at various web sites, which you can find with a suitable search.
The format of the files at the http or ftp site corresponding to two URLs must correspond to the formats expected by the usual install procedures of the operating system that you are installing. (Paid support.)
Analogous boot lines are used for type M operating systems.
Q. Why does it say “Not enough free space for auto-allocating”?
Mandrake Linux 10.1 may yield this error on the A3-DP server. If this happens, you will need to partition the installation disk manually, as described below, and you will not be able to use software RAID. (Paid support.)
If you wish to use Mandrake Linux, you should sign up for one of the other servers, and not for the A3-DP server.
Q. Why aren't mirrored RAID pairs identical in size?
Our observations indicate that the installer for Mandrake Linux 10 and 10.1 partitions disks erratically when software RAID is in use. After installation is completed, use the “df” and “fdisk -l” commands to display the partition sizes. One of the two partitions of a mirrored pair (as listed by “fdisk -l”) might be too big, thus wasting some disk space. However, the filesystem itself (as listed by “df”) will usually have the desired size. Other than some waste of disk space, we have not noticed any adverse effects.
Q. Why do I see strange perl assertion errors during installation?
When you begin installing Mandrake Linux 10.1, the installer will ask you to specify which installation CDs are available. Make sure CDs 1, 2, and 3 are checked, and CD 4 is unchecked. Installation should then complete normally.
If you leave CD 4 checked, the installer will attempt to install KDE packages that are not available, and will yield error messages similar to “perl: ./rpmio_internal.h:497: c2f: Assertion `fd && fd->magic == 0x04463138' failed”.
Q. Why does it say “Cannot read printrc file” during installation?
This is normal and does not indicate a problem.
Q.Why doesn't my server reboot after installation completes?
On some hardware platforms, after operating installation completes, the installer software may fail to successfully reset the system hardware. You will see messages such as “unmounting filesystems” and “restarting system” and then nothing.
Wait about two minutes before you conclude that the reboot did not occur. Then go to the Server Configuration web page (see Activity Links) and power-cycle your server. It should then reboot normally.
Q. Why does the serial console not let me log in?
If you install the operating system in uninterruptible or interactive mode, the server should have been automatically configured to allow logins, including root logins, on the serial console.
If you have installed the operating system in manual mode, the operating system in its default configuration might not have a “getty” process active on the serial console. If this is the case, you should be able to correct it by first logging into the server as root using “ssh”.
While logged in as root, edit the file “/etc/inittab” and append to it a line like this:
co:2345:respawn:/sbin/agetty ttyS0 9600 vt100
Then while still logged in as root, use the following command to make the operating system (really the “init” process) reread the “/etc/inittab” file:
# telinit q
Additionally, you may need to explicitly permit permit root logins on the serial console, otherwise only non-root users will be able to log in. If the file does not exist, then nothing more should be needed. If the file “/etc/securetty” does exist, then append to it this device name on a separate line:
ttyS0
Also see the next FAQ entry below.
Q. In what ways does installation modify the operating system defaults?
Our uninteruptible and interactive install options (but not the manual install option) will make some convenience changes as described below. (Not every change applies to every operating system.) Whenever an existing file is changed, the original is left behind with a .ORIG extension.
Q. Why does my server have too many CPUs?
If your server has a CPU with hyperthreading, this means the CPU can sometimes do two tasks at a time. During operating system installation, a multiprocessing kernel will be installed. The output from the “top” command may show two cpus called “cpu0” and “cpu1”. The two CPUs are virtual CPUs, not real CPUs. The performance gain due to this, if any, will likely vary with the job mix on your server. Our performance numbers (see the rate charts elsewhere) will give you some idea of the performance benefit of hyperthreading.
Q. My does my server with SATA disks report having SCSI disks?
The Linux kernel uses an SATA-to-SCSI translation layer to make the SATA disks look like SCSI disks. The rest of the kernel then uses the SCSI driver to communicate with the disks. This is normal and is nothing to worry about.
Q. Why is there an “a2iroot” account on my server?
The optional “a2iroot” facility lets us conveniently log into your server as root. See the item “Allow a2iroot access” on our Server Configuration web page (see live demo) and see its online help. Also see “In what ways does installation modify the operating system defaults?” elsewhere on this page..
Q. I did a manual install; why can't I see the kernel boot messages from the serial console?
A. You will need to configure the machine to use the serial console. Download the file “ftp://servers.rahul.net/setup/setup.serialconsole”, and while logged in as root, run it with this command:
# sh setup.serialconsole
This should configure the serial console for you. After the next reboot, you should see the kernel messages on the serial console at boot time. XXX: does this work for mdk?
If this doesn't work, you may refer to the Remote Serial Console HOWTO and the Linux Serial Console HOWTO, or you may re-install the operating system choosing an uninterruptible or interactive install
Q. Why do the boot-time messages say there is a “bug”?
The Linux kernel contains many special cases to properly drive a wide variety of hardware devices. If any such special case is detected at boot time, the kernel will usually report a “bug” at boot time. Most common hardware triggers one or more of these special cases causes a “bug”' message at boot time.
Q. I'm seeing very few boot-time messages; where is all the kernel output?
Our uninterruptible and interactive install procedures install the boot loader so it will feed the argument “quiet” to the kernel at boot time. This suppresses the less important boot-time messages. To see all of them, simply remove the argument “quiet” from the file /boot/grub/grub.conf or /boot/grub/menu.lst (whichever is in use on your system) on each line that begins with the word “kernel”. Note: if you make the wrong edits on grub.conf or menu.lst, you may end up with a system that no longer boots.
Q. What sort of technical support can I expect from you?
We provide both free and paid support, as follows.
Free support is always available related to billing issues.
Tehnical support, not related to billing issues, is available as below.
If you have completed our New Server Procedure (see Getting Started), then you are eligible for both free and paid technical support, as described below. Only the individual who has completed the New Server Procedure is eligibible for free technical support.
If you have not completed our New Server Procedure, then only paid technical support is available to you, except that we provide free technical support solely to help you complete the New Server Procedure.
Free technical support. We respond to specific problem reports about the services for which you are paying us. These services include: providing you with functioning hardware, and providing you with operating system installation, rescue, and upgrade procedures, that you can control remotely. If any of the features that we advertise do not work as described, we will help you resolve the problem, with the exceptions that: (a) we do not provide any free support to help you overcome problems inherent in the operating system or its installation software and (b) to the extent that any questions that you ask us duplicate questions that are already answered in the documents on this web site, the time that we spend answering your questions will be billed as paid support time.
In some places on this web site, we describe specific other types of free support that we offer.
In all cases, to get free technical support as described above, we require you to follow our published procedures wherever these are provided. For example, elsewhere in this FAQ document we describe how to report hardware errors and to run diagnostic tests. If you do not follow our published procedures, then free technical support is not be available to you, but we will usually still provide you with paid technical support.
Also, wherever on this web site you see the phrase “paid support”, it means that all technical support about that issue is available only as paid technical support and not free technical support. (See Miscellaneous Fees.)
If you have specific needs and want to make sure they are met by one of our dedicated servers, please contact us before signing up for one. By prior arrangement, we may be able to offer you a limited amount of free or low-cost technical support to help you meet those needs.
Paid technical support. All technical support that you need with regard to the operating system and other software on your server, after installation has been succesfully completed, is available only as paid technical support.
Although we provide some convenience features that become useful after the completion of operating system installation (e.g., easy upgrades via our local upgrade server), those features too are eligible only for paid technical support, not free technical support. We always gladly accept problem reports about these features at no charge to you, and we will fix any problems at our own discretion. But if you need any assistance from us with these convenience features beyond simply reporting problems to us, such assistance comes only as paid technical support.
Q. How can I recover a damaged boot block or partition table?
Two methods are described below. You will probably need to use only one of these methods. You are advised to practise using these methods well in advance of when you actually need them
Both methods use the program “testdisk” to search for partitions on disk and rebuild the partition table. Although “testdiskc” is a very smart program, there is no assurance that it will always produce the right results. It is easy to accidentally write the wrong partition table on disk. We advise practising with “testdiskc” on a machine at your home or office that does not have important data on it, and learning to expertly use this program, before invoking it on your dedicated server.
We have tested this procedure on the RedHat 9 and Fedora Core 2 platforms and it worked to our satisfaction. Recovering a damaged boot block or partition table is a delicate operation that is not guaranteed to always work.
Method 1. From the PLD Rescue CD, if supported by your server. Using serial console access, reboot the machine, select the Utilities Menu, then select the PLD Rescue CD. When the PLD Rescue CD has finished booting, it will give you a login prompt. Log in as root. It will not ask for a password. Then set your terminal height to the correct number of rows (e.g.: “stty rows 30”). Then invoke the “testdisk” program , giving it the device name for the boot disk; e.g.: “testdisk /dev/hda”). Now you may search for partition information and attempt to rebuild the partition table, keeping in mind all the warnings given above.
For an actual example of how we did this, please see our Demo 2.
Method 2. From the rescue mode of your operating system. Following are instructions suitable for type R operating systems. Using serial console access, reboot the machine. From the Initial Menu, go to the Utilities Menu, then select the rescue mode for your operating system. (If a warm reboot does not give you the Initial Menu, see “Why don't I see the Initial Menu when I reboot my server?.”)
Once the rescue mode has finished loading, go into chroot mode, and fetch a copy of testdisk, as shown below. Everything shown below, except for the root prompt #, will be typed by you.
# chroot /mnt/sysimage # cd /tmp # wget ftp://servers.rahul.net/util/testdisk_static # chmod +x testdisk_static
If for some reason the rescue mode has trouble resolving host names to IP addresses, go to another window, manually do a DNS query to resolve “servers.rahul.net” to an IP address, and use that IP address in the “wget” command above.
At this point you should have in the current directory an executable program called “testdisk_static”, which is a copy of “testdisk” linked statically, so it is not dependent on any dynamic libraries. Now set your terminal length (assumed to be 30 below, but you should use the correct value), and set your terminal type to “bterm”, and then invoke “testdisk” on the desired disk:
# stty rows 30 # TERM=bterm # export TERM # ./testdisk_static /dev/hda
At this point “testdisk” should be running and you may attempt to rebuild the partition table, keeping in mind all the warnings given above.
You may also need to reinstall the boot block, using the “grub-install” program which should be already available in both cases above.
For more information about the software mentioned above, please see the appropriate web site.
The methods described above have been tested by us with some, but not all, machine and operating system configurations. You should feel free to try other tools. Once you have booted into the rescue mode of your operating system, you should be able to download other programs (preferably statically-linked ones) from the Internet and run them while in rescue mode.
Q. How can I reset a forgotten root password?
Two methods are described below. You need use only one of these methods. You are advised to practise using these methods well in advance of when you actually need them.
The specific examples given below are for Fedora Core 2. Other operating systems are analogous though not necessarily identical.
Method 1. Using serial console access, reboot the machine (by power-cycling it, if you can't log in as root). Select the Utilities Menu, then select the Install/Rescue menu for your operating system, and then select the rescue mode. The sequence of menus will be:
Utilities Menu => Fedora Core 2 Install/Rescue => fc2 rescue mode
If asked to choose a netwok interface, choose eth0. When asked for network information, select the default, which should be “Use dynamic IP configuration (BOOTP/DHCP)”. When prompted, select “Continue”. When prompted, select “OK”. When you get a root prompt, enter the command “chroot /mnt/sysimage”. At the next root prompt, you will be logged in, with all filesystems mounted normally. Now use the “passwd” command (without any arguments) to set a new root password. The last few steps are shown below. The portions that you would type will appear as underlined on many browsers. The password that you would enter is shown as “******”, but it does not echo on the screen.
# chroot /mnt/sysimage sh-2.05b# passwd Changing password for user root. New UNIX password:****** Retype new UNIX password:****** passwd(pam_unix)[337]: password changed for root passwd: all authentication tokens updated successfully. sh-2.05b# exit exit # exit sending termination signals...done (... the system reboots ...)
Method 2. By going into single-user mode as follows. Reboot the machine (by power-cycling it, if you can't log in as root). Let it reach the menu where the kernel is about to be booted. Before the timeout expires, hit the “e” key. This gives you a grub edit menu. Select the “kernel” line and hit the “e” key again. You will now be editing this line, and the cursor will be at the end of the line, just after the words “console=ttyS0,9600n8”. Type a space and then the word “single”. The line will now look like this:
kernel ... console=ttyS0,9600n8 single
Now hit the ENTER key. Then hit the “b” key. The machine will boot into single-user mode and will give you a root prompt. Use the “passwd” command to set a new password. Then reboot.
Note that Method 2 will work only if you have not enabled boot-time security. For example, if there is a “/etc/securetty” file and it does not list “console” as one of the entries, then boot-time security is enable, and the single-user mode will ask you for the root password before giving you the root prompt. This leaves you with a chicken-and-egg situation. Then you will need to use Method 1.
You may also ask us to reset a lost root password for you. See Miscellaneous Fees.
Q. When I reboot my server, why can't I access the Utilities Menu?
Depending on the specific hardware, you might or might not see the Utilities Menu each time the server is rebooted (e.g., with the “reboot” command). If this happens, here are two methods of accessing the Utilities Menu.
Q. Why is the Fedora Core 2 kernel newer than the original?
The Fedora Core 2 installation files that we supply have been updated with a newer kernel that was distributed after the initial Fedora Core 2 release. This has been done in order to properly support SATA disks.
Q. How can I update the operating system?
The word “update” as used here means bringing an operating system up to date with the latest fixes, while keeping its name and revision number the same. Updating Fedora Core 2, for example, still leaves it as Fedora Core 2. This is not the same as an “upgrade”, which converts an operating system release into a later release.
You are required to use reasonable diligence to keep your server updated with all applicable officially-available security fixes.
For each operating system available for installation from our menus, after an uninterruptible or interactive installation, the operating system may be easily updated with one or two easy commands. The table below tells you how to update your server depending on the operating system that has been installed. First log into your server as root, and then invoke the command given in the “update command” column.
| Operating System | Update Command | Notes |
|---|---|---|
| RedHat 9 | yum -y update | Console |
| Fedora Core 2 | yum -y update | Console |
| Fedora Core 3 | yum -y update | Console |
| CentOS 3.3 | yum -y update | Console |
| Mandrake 10 | MandrakeUpdate | Window |
| Mandrake 10.1 | MandrakeUpdate | Window |
"Console" means that the program runs in a console (i.e., terminal) window and generates no graphical output. "Window" means the program requires an X Windows environment and may only be run with output sent to a machine that has a graphical display and an active X server process. In each case, the program must be invoked as root.
If the “MandrakeUpdate” command is not found, then install it as follows. Go into the directory /root and use the following command:
# sh setup.rpmdrake
(If the above is not found, try running “setup.rpmdrake.done” instead.
This will attempt to re-install the rpm packages containing the “MandrakeUpdate” command.
Then to do an update, invoke the command by its full path “/usr/bin/MandrakeUpdate”.
Q. How can I upgrade the operating system?
The word “upgrade” as used here means converting an operating system release to a later operating system release that has a different name and/or a different release number. For example, if you began with RedHat 9 and ended up with Fedora Core 2, this would be an upgrade. This is not the same as an “update”, which means bringing a certain operating system up to date with the latest fixes, while keeping its name and revision number the same.
Our menus give you the opportunity to use the upgrade functionality built into the operating system to which you wish to upgrade. The success of an upgrade may greatly depend upon the operating systems involved, which specific packages exist on the server before the upgrade, and how you are using the server. A fresh reinstall of an operating system is often preferable to doing an upgrade.
In our experiments, we were able to use our menus to upgrade as follows.
An alternative to using the upgrade functionality built into the operating system is to attempt an upgrade using “yum”. Below are some links to web pages (not connected with us) that discuss advantagers and disadvantages of using “yum” in this way.
Q. When I type “yum update”, why does it say “command not found”?
The “yum” program is already included with Fedora Core 2, Fedora Core 3, and Centos 3.3, but not with RedHat 9 and Mandrake 10. Our uninterruptible and interactive install choices will install yum for you on RedHat 9 if it is not already present. If you selected packages fewer than our default, then there might be some packages missing that are required to install and use yum. This will cause yum to not get installed in the case of RedHat 9. You can try manually downloading and installing yum, or you can do an interactive operating system reinstall and select “Everything”, or you can do an uninterruptible operating system reinstall, which automtically installs “Everything”.
If none of the above suggestions successfully installs yum for you, simply download yum yourself from a suitable site on the Internet and install it yourself. (Paid support.)
Yum is not currently supported by Mandrake 10. Use “MandrakeUpdate” as described above.
Q. I did “yum update” or “MandrakeUpdate”, and now my system is having problems; can you fix this for me?
The updates provided by our update server are merely mirrored from official sources, and not generated by us.
Also, note that our update server fetches updated packages from the official distribution sites once each night. If an official distribution site is not accessible at that time, our update server will try again the next night. It is therefore possible for our update server to lag behind the official distribution sites by as many days as it takes to make contact and fetch the updates.
In case of any problems with applying these updates, you may consult information sources on the Internet, such as the Fedora Legacy Project (for RedHat Linux) the Fedora Project (for Fedora Core), or the cAos Community Linux group (for Centos).
The updates installed by “MandrakeUpdate” come directly from mirors that provide updates issued by the operating system supplier. In case of any problems with applying these updates, you may consult the web site for Mandrake Linux .
Q. How can I use X Windows on the server?
You will need to direct the display so it goes across the network to your own graphical terminal. (Paid support.) Note that the X Windows programs that you wish to run must be available on the server. Most X Windows programs should already be installed if you did an uninterruptible operating system install.
Q. What if there are hardware problems?
If you suspect a hardware problem, you have an obligation to make an initial diagnosis as described below.
If we cannot quickly identify and fix the hardware problem that you observe, we will offer to switch your service to a replacement machine. We will allow 48 hours of overlap between the old and replacement machines at no charge to you. If you wish to keep the old machine online for longer than that, you will be billed for both the old machine and the replacement machine for the additional overlap time.
Customers using remote servers often fail to take steps to back up their data. If there is ever a disk failure, they experience data loss. By providing dual disk drives on most of our servers, we encourage you to have some sort of backup mechanism in place.
You can configure two disk drives into a RAID configuration with mirrored partitions (by checking the Enable level 1 software RAID button on the Server Configuration screen).. This will substantially decrease the chances of loss of data due to the failure of a single disk.
Or you may choose not to use RAID, but instead, keep spare partitions on one disk onto which you periodically make backup copies of valuable data from the other disk.
Note: The mechanisms described above are better than doing nothing. The best backup mechanism, however, is to periodically copy valuable data to some other location not near your server. For example, you can download important data from the server to your home or office machine and burn it to CD-ROM.
ECC means “error correction code”. A server that is expected to process data reliably 24 hours a day, 7 days a week, should use ECC memory. ECC memory automatically corrects single-bit errors, so that the server can continue operation uninterrupted. Our use of ECC memory follows industry practice in the area of high-quality, reliable computer systems. It eliminates one possible cause of mysterious data corruption and machine crashes. “The PC Guide” has a useful discussion on memory errors.
Q. How can I check the server's hardware configuration?
Q. How can I check fan speed, voltages, and temperatures on my server?
This can be done on most but not all servers, as specified on our rate charts. If your server is a Supermicro model, get the file “ftp://servers.rahul.net/setup/setup.supermicro”, and run it while logged in as root with the command “sh setup.supermicro”. It will install two monitoring programs on your server. Run the program “/usr/sbin/sdt” for non-graphical text output. Run the program “/usr/sbin/superodoctor” for X Windows output.
CPU temperature will vary with the load on the CPU. Under intensive processing conditions, the CPU temperature may rise 10 to 15 degrees Celsius higher than when the machine is idle.
Our servers do not use any CPU fans (see more about this below), so the CPU fan speed will be displayed as zero. If the server has fewer chassis fans than the program expects, some chassis fan speeds may also be displayed as zero.
Temperatures are displayed as two numbers separated by a slash. The first number is in Celsius, and the second number is in Farenheit.
The stated maximum and minimum values are only estimates. It is possible for an actual measurement to be outside the stated maximum and minimum values. This does not necessarily indicate a hardware problem.
Q. Why don't your servers use CPU fans?
CPU fans became popular some years ago when many cheap, poorly-built PCs came on the market. A vigorous “aftermarket” in cooling devices developed. Among these cooling devices were fans to cool the CPU in poorly-designed machines.
At a typical speed of 4,000 rpm, a CPU fan will rotate about two billion times during one year of operation. If the wheels of your car rotated two billion times, you could drive far enough to get to the moon and back—not just once, but seven times. The hotter the fan bearings run for those seven round trips to the moon each year, the more quickly the lubricant will dry out, and the more likely it is that the bearings will jam. Fans mounted on the CPU run hotter than fans not mounted on the CPU, and will have a higher rate of bearing failure.
Our servers use cooling fans not mounted on the CPU. No matter how hot the CPU runs—and today's CPUs are designed to run quite hot when doing intensive processing—the fan bearings run cool.
Q. What's the difference between your and everybody else's servers?
Q. Who owns the dedicated server?
Although we refer to the dedicated server as “your” server, you are really buying a service, not a machine. We allocate a server to you for the duration of your service.
Q. Can I really use 600 gigabytes of bandwidth?
In peak months, you may use up to the amount of transfer volume stated in our price list. The transfer volume is calculated by adding up both the incoming and outgoing transfer volume. This policy assumes that your server will have a smooth transfer characteristic, and with peak transfer rates staying below 3 megabits per second. The monthly limit is intended to protect you from unpredictable usage and unexpected charges, and is not intended to be the sustained transfer volume used by you over a long period of time. Your server will be on a shared network, and the monthly limit assumes that only a small fraction of servers on the network will operate close to the limit in any given time period.
If your monthly volume is low most of the time, but peaks up to the limit only rarely when something unexpected happens, then your usage is within our guidelines. Unexpected things include, for example, being mentioned in a radio or TV show that causes a sudden onrush of downloads.
If we observe unusual bandwidth beyond the expectations of the description above, we will attempt to contact you to discuss the situation, and we may quote you a monthly rate other than the standard rate. In unusual cases (e.g., peaks of higher than 5 Mbps, or in other situations that adversely affect our network), or if we are unable to contact you reasonably quickly, we may add bandwidth throttling, or take your server offline.
If your monthly volume is high because you set things up that way, then please see below.
Q. What are my high-bandwidth options?
Below are some high-bandwidth options. In each case:
Here are the high-bandwidth options.