I am wondering if (in any state of development) there exists a filesystem that can inherit files from another filesystem, but changes made on that mount point would only affect that mount point.
This would be sort of like a distributed filesystem that only synchronizes inward and never outward, and only if the lack of changes permit that action.
In the past I have used Samba with Active Directory to allow network users to log into Linux machines. Today I'm maintaining a smaller network comprised only of Linux hosts, but I would still like to centralize authentication. I have looked into setting up a Samba host as a Domain Controller, but I don't exactly see the point of doing this when there are no Windows hosts. I've looked into NIS but it seems like it might be a bit complicated for my taste.
It is important that network-authenticated users share the same UIDs because there will be some shared disk resources involved as well.
For what it's worth the majority of my network nodes are OpenVZ containers, though there are some outliers (laptops, workstations, and a couple dedicated servers).
Is it feasible to create a single "master" OpenVZ guest who would only be used for package management, and use something like mount --bind on several other OpenVZ guests sort of trick them into using the environment installed by the master guest?
The point of this would be so that users can maintain their own containers, and yet stay in sync with the master development environment, so they'll always have the latest & greatest requirements without worrying too much about system administration. If they need to install their own packages, could put them in /opt, or /usr/local (or set a path to their home directory)?
To rephrase, I would like several (developer's, for example) OpenVZ guests whose /bin, /usr (and so on...) actually refer to the same disk location as that of a master OpenVZ guest who can be started up to install and update common packages for the environment to be shared by all of this group of OpenVZ guests.
For what it's worth, we're running Debian 6.
Edit:
I have tried mounting (bind, and readonly) /bin, /lib, /sbin, /usr in this fashion and it refuses to start the containers stating that files are already mounted or otherwise in use:
Starting container ...
vzquota : (error) Quota on syscall for id 1102: Device or resource busy
vzquota : (error) Possible reasons:
vzquota : (error) - Container's root is already mounted
vzquota : (error) - there are opened files inside Container's private area
vzquota : (error) - your current working directory is inside Container's
vzquota : (error) private area
vzquota : (error) Currently used file(s):
/var/lib/vz/private/1102/sbin
/var/lib/vz/private/1102/usr
/var/lib/vz/private/1102/lib
/var/lib/vz/private/1102/bin
vzquota on failed [3]
If I unmount these four volumes, and start the guest, and then mount them after the guest has started, the guest never sees them mounted.
I've migrated a server from vmware to openvz, both are running on the same physical host (I plan to eliminate vmware). If I give the new openvz container a new IP anybody on the LAN and outside network can get to it with no problem, but as soon as I change its IP address to the original IP (that the vmware guest used to have), only the host server and its contained virtual machines can get to it.
I have made sure to change the IP on the old virtual machine.
Is there a device that I can plug SATA disks into to erase the MBR? I know this could be done with a computer, but I'd rather have something compact, not unlike a disk duplicator. In fact I've even thought about using a disk duplicator to do this, but that would mean having to wait as long as it takes to duplicate the master disk, and since I can't find a 512-byte sata disk (edit, I haven't actually looked for one), that would be a lot of time wasted.
Edit:
The reason I'd like something is that I'm doing a LOT of these disks, and the reason I'm doing this is so that I can assemble the hardware, and have them PXE boot to install the operating system. Some of these disks may be refurbished, so if there's a bootable disk installed, the BIOS may not automatically hit the PXE ROM.
I'm setting up an installer using both PXE and USB, and it works well, however I need to add a few different options to the boot menu to use different preseed files. This is for different situations, like whether or not to set up software raid, and with how many disks, etc.
So ideally, I'd like to roll one initrd image containing several preseed files, and specify which one to use with a kernel argument, as detailed in the manual. However, this isn't working. It seems that no matter how I specify preseed/file= , the installer will only look for /preseed.cfg .
Here is an example of my pxelinux.cfg/default file (a very similar config exists for syslinux as well):
DISPLAY boot.txt
DEFAULT install_raid_1disk
LABEL install_raid_1disk
kernel installer/2b/linux
append vga=normal initrd=/installer/2b/initrd.gz preseed/file=/preseed-net-raid-1disk.cfg --
LABEL install_raid_2disks
kernel installer/2b/linux
append vga=normal initrd=/installer/2b/initrd.gz preseed/file=/preseed-net-raid-2disks.cfg --
LABEL install_noraid
kernel installer/2b/linux
append vga=normal initrd=/installer/2b/initrd.gz preseed/file=/preseed-net-noraid.cfg --
PROMPT 1
TIMEOUT 20
Some Background...
I think this should be a general linux kernel (or kernel module) related question, though I'm facing the issue while preseeding debian for an automated installation.
Every time I boot the installer from a usb disk, the usb disk turns out to be /dev/sda, but the preseed file I'm using assumes the first hard disk (sata in this case) should be /dev/sda. I'm using the same system for a pxe network installation, and I would prefer to keep them consistent. I also fear that if I wrote a preseed config that assumed the first disk was sdb instead, that for some reason it might not be at some point.
I've tried writing udev rules that changes the device name of the usb disk to "INSTALLMEDIA" which seems to work, however it does not cause the SATA disk to start at sda, they still start at sdb.
My questions...
What determines the order that disks are discovered and named?
What, if anything, can I do to control this?
If I cannot do anything to control it, would it be safe to assume when booting from USB that the USB disk can be more or less guaranteed to be sda and the first SATA disk would always be sdb?
This is disgusting to me, but it works:
echo -e "n\np\n1\n\n\nt\nc\na\n1\nw" | fdisk /dev/sdb
I would like to know a more prudent or elegant way to accomplish this. The entire boot record and partition table is being zeroed prior to this task, so we can trust that partition 1 is available.
Let's assume I have a several gigabyte tar file, but I also happen to know that the very last file written to the archive is something important that I need. Since tar files are appended sequentially, is there a way I can make tar read into the archive from the end to find this file, instead of starting from the beginning and reading over gigabytes of irrelevant data?
Is it possible to set an arbitrary path within a tar archive to place files which may not exist on your original filesystem? This differs from the -C switch in that I may want to take a back up of /etc/, but place it in the archive as /configurations/etc/.
I'm setting up an environment to install many Lenny machines at once and I need to avoid any dialog with the process, corner cases aside. One thing that consistently troubles me is that after my preseeded disk recipe creates Linux RAID-1 devices, the installer warns me about how the kernel could not re-read the partition table on each MD device. When I do a manual installation and try to use debconf-get-selections to preseed another installation, it seems to ignore the whole preseed file altogether. But reading through those selections, I couldn't find anything that seemed to suggest it would select "OK" on this extraneous warning.
I actually have two questions here regarding the preseed.cfg file. One being how to calculate how much disk space to use for my scheme during the installation (for when I might be netbooting to install various sizes of hard disks). The other being how to set up linux software RAID from the preseed.cfg file.
Rather than automatically using the whole disk, I would like to set it up to create 2GB MD device for use as swap, a 3GB MD device for use by ext3 at /mnt/utility, and all the rest of the disk space as the first partition, also an MD device for use as ext3 mounted at /. Is there a way to specify sizes for some partitions and then allocate "everything else" to one partition?
I'm having a hard time finding linux RAID examples for debian's preseed.cfg--in fact I have no idea what that would look like.
I would like to set up Exim4 on Debian as an authenticated smart host, which behaves not terribly different from gmail (or google apps).
I want it to accept mail via TLS, require username and password credentials, and always rewrite the sender as the authenticated user.
My understanding from the manuals is that -r sets the return path and -f sets the sender's email address, however I'm finding that they both set both, and whichever one comes latest takes the priority.
Since this binary is actually provided by Exim4, should I assume all bets are off regarding the behavior of classic sendmail?
In my situation I need to ensure one particular filesystem is mounted before a few subsequent filesystems should be. In this case, for example, I want to mount a USB disk before bind mounting some other directories on it.
I am comfortable enough to write an init script to accomplish this, however before doing that I would like to know if there might be a more prudent or common solution to this problem--I'm skeptical that this problem is a rare one.
I know this isn't the purpose of TFTP, but I'm working in an environment where a lot of different types of devices pull provisioning info from a TFTP server. What I'm developing is a provisioning system that tracks and maintains device configurations, and I would like to have the requested files generated on the fly, much like you could do with any web application.
Yes some of these devices can support HTTP for provisioning, but not all of them do, and we want things to be consistent.
Are there any TFTP daemons that can provide something analogous to CGI?
I'm helping out with a project for an embedded device that only has USB and no serial. This device is running Linux.
These days, when we need to connect to a serial port on a device we typically use a USB to serial adapter (on something like a phone system or a load balancing device, etc). I would like to know if it is possible to have the host device behave as though it were a serial adapter, thus removing the need for one. Given the nature of USB, is this approach even necessary?
To recap, I would like to be able to connect a single A-to-A USB cable from my workstation (be it windows or linux) to this device, for the purpose of administration (especially initial setup), using minicom, putty or hyperterminal.
Thanks
Edit
Okay, I'm admitting defeat on this idea. That said, I'm thinking that it would be possible to accomplish this goal using two usb-serial adapters and one serial cable?
The problems with this approach would be:
The case I have is a directory that stores phone call recordings throughout the day. The root filesystem is running out of space so we added an external hard drive. This server may not be taken out of production for any reason, so what I would like to do is mount the new disk over the same recordings directory, and then after all of the calls in progress have finished I would copy them from the old filesystem into the new one.
I will be able to determine when the calls in progress during this transition have been finished up.
My question is how can I access the files beneath /var/x/y/recordings/ after I mount a new disk onto /var/x/y/recordings/?
I believe an open file should be able to finish being written to and eventually closed after a new FS is mounted over its parent directory. Is this incorrect?
When I search around for example RAID1 installations, it seems that admins tend to place their swap partition under RAID1.
To me, it is more intuitive to have two disks, each with a large partition for use by RAID1 and the system partition and with a smaller partition for swap, outside of the RAID array.
What is the worst-case-scenario if I lost a disk and effectively half of my swap space while the system is running?
Should I expect to see a performance increase or decrease while mirroring a swap volume vs having two separate swap volumes outside of RAID?
If swap should be mirrored, does it make more sense to give swap its own RAID1 array, or does it make more sense to partition one big RAID1 array with LVM?
(Note/4. I'm not sure if mdX can be partitioned without LVM, but the debian installer leads me to believe that it cannot)
Sparing most details/reasons, I have a situation in which I'm cloning drives for mass production. Some of them are being installed in servers that only need one disk (in that case this is a non-issue), but some are being installed in servers that will use 2 of them in RAID1.
In another case we're building a tool to assist the admin in replacing a faulty disk by ensuring the new disk is large enough and then copying the MBR and Partition Table, and then adding it to the RAID1. When this happens LILO complains about duplicate volume IDs (which makes sense).
This is why I would like to generate new volume IDs for the partition being added to the RAID1 array. Does it make sense to use sfdisk to rewrite a disks partition table, or is there a more straightforward command/technique to create a new volume id for an existing volume?
