emerge over chrooted nfs share?

[peter4]

I've been wondering: is it possible to export one (slow) computer's whole filesystem via nfs, chroot into it from a more powerful computer and do the heavy lifting from there (emerge chromium)?

I've tried setting it up, but had no success so far (it's the first time I'm dealing with nfs). I'm stuck at a point where I can mount the nfs share, but I can't access filesystems mounted there, like /proc or /sys. They appear as empty dirs. I tried "nohide" option, with no effect.

[Hu]

Why not use emerge --buildpkg on the fast computer and emerge --usepkgonly on the slow computer?

[Jacekalex]

Discc?

[bbgermany]

I've been wondering: is it possible to export one (slow) computer's whole filesystem via nfs, chroot into it from a more powerful computer and do the heavy lifting from there (emerge chromium)?

I've tried setting it up, but had no success so far (it's the first time I'm dealing with nfs). I'm stuck at a point where I can mount the nfs share, but I can't access filesystems mounted there, like /proc or /sys. They appear as empty dirs. I tried "nohide" option, with no effect.

since you will need the /proc and /sys of the compiling system (the fast system), you should try mount via bind the local /sys and and /proc this way, before chroot into to nfs system:

Code: Select all

mount -o bind /proc /mnt/nfs/proc
mount -o bind /sys /mnt/nfs/bind

in some cases you need the /dev as well. afterwards just do as stated in the handbook and everything should work.

bb

[Chewi]

I won't go into full detail just now but I do this with several machines using this script I wrote. Create a symlink to it like chroot-hostname.

Code: Select all

#!/bin/sh

HOST=${0##*/}
HOST=${HOST#*-}

mkdir -p --mode=0755 /mnt/${HOST}

mount -t nfs -o rw,intr,noatime,actimeo=60,vers=4,fsc ${HOST}:/ /mnt/${HOST}
mount --bind /dev /mnt/${HOST}/dev
mount --bind /dev/shm /mnt/${HOST}/dev/shm
mount --bind /proc /mnt/${HOST}/proc
mount --bind /sys /mnt/${HOST}/sys
mount --bind /usr/portage /mnt/${HOST}/usr/portage
mount --bind /usr/local/portage /mnt/${HOST}/usr/local/portage
mount --bind /var/tmp/portage /mnt/${HOST}/var/tmp/portage

env -i - HOME="/root" TERM="$TERM" chroot /mnt/${HOST} /bin/bash -l

umount /mnt/${HOST}/dev/shm
umount /mnt/${HOST}/dev
umount /mnt/${HOST}/proc
umount /mnt/${HOST}/sys
umount /mnt/${HOST}/usr/portage
umount /mnt/${HOST}/usr/local/portage
umount /mnt/${HOST}/var/tmp/portage
umount /mnt/${HOST}

NFSv4 can be quite confusing, even if you've used NFSv3 before. I also sometimes get a very long delay before the bash prompt actually appears, even though no network traffic is passing during that time. I haven't been able to figure out why.

[Hu]

Code: Select all

mount --bind /var/tmp/portage /mnt/${HOST}/var/tmp/portage

Although I understand you do not want to send Portage temporary files over the network during compilation, binding your local /var/tmp/portage is a bit unsafe. If you are merging the same package on multiple machines at once, they will all be sharing a work area, which could cause unwanted behavior. You could avoid this by creating /var/tmp/portage/$HOST and mounting it into the chroot as /var/tmp/portage.

I also sometimes get a very long delay before the bash prompt actually appears, even though no network traffic is passing during that time. I haven't been able to figure out why.

Have you checked which process is blocked during this time? Specifically, is it stalled waiting for a mount to finish or is it stalled waiting for the chroot'd bash to initialize? You can use set -x at the top of your script to print each line before it executes, so you can see how far it gets before the stall occurs.

[Chewi]

Although I understand you do not want to send Portage temporary files over the network during compilation, binding your local /var/tmp/portage is a bit unsafe. If you are merging the same package on multiple machines at once, they will all be sharing a work area, which could cause unwanted behavior. You could avoid this by creating /var/tmp/portage/$HOST and mounting it into the chroot as /var/tmp/portage.

Good point, I hadn't thought of that. I don't usually build on more than one at once but as it happens, I was today. Chances of a conflict are low but yeah, it could happen and wouldn't be pretty.

Have you checked which process is blocked during this time? Specifically, is it stalled waiting for a mount to finish or is it stalled waiting for the chroot'd bash to initialize? You can use set -x at the top of your script to print each line before it executes, so you can see how far it gets before the stall occurs.

It's not waiting for the NFS mount because if I do ls /mnt/hostname during the delay, it returns with the remote contents immediately. It's something to do with chroot and/or bash.

It's not the cause (it happened before) but I forgot to mention that I am using cachefilesd to help speed things up a bit though I'm not sure how effective it really is. I still get a lot of network traffic even for repetitive operations.

[Hu]

Since you know the specific hung command, you could modify your script to run that command under strace with output saved to a file. Try strace -ff -tt -o /tmp/nfschroot.strace to get per-process traces, with timestamps. Once the hang ends, you can examine the traces to see which process(es) were blocked and what system call they made right before blocking.

[Joseph_sys]

I won't go into full detail just now but I do this with several machines using this script I wrote. Create a symlink to it like chroot-hostname.

Code: Select all

#!/bin/sh

HOST=${0##*/}
HOST=${HOST#*-}

mkdir -p --mode=0755 /mnt/${HOST}

mount -t nfs -o rw,intr,noatime,actimeo=60,vers=4,fsc ${HOST}:/ /mnt/${HOST}
mount --bind /dev /mnt/${HOST}/dev
mount --bind /dev/shm /mnt/${HOST}/dev/shm
mount --bind /proc /mnt/${HOST}/proc
mount --bind /sys /mnt/${HOST}/sys
mount --bind /usr/portage /mnt/${HOST}/usr/portage
mount --bind /usr/local/portage /mnt/${HOST}/usr/local/portage
mount --bind /var/tmp/portage /mnt/${HOST}/var/tmp/portage

env -i - HOME="/root" TERM="$TERM" chroot /mnt/${HOST} /bin/bash -l

umount /mnt/${HOST}/dev/shm
umount /mnt/${HOST}/dev
umount /mnt/${HOST}/proc
umount /mnt/${HOST}/sys
umount /mnt/${HOST}/usr/portage
umount /mnt/${HOST}/usr/local/portage
umount /mnt/${HOST}/var/tmp/portage
umount /mnt/${HOST}

NFSv4 can be quite confusing, even if you've used NFSv3 before. I also sometimes get a very long delay before the bash prompt actually appears, even though no network traffic is passing during that time. I haven't been able to figure out why.

Can you explain me what these two line do in your script?

Code: Select all

#!/bin/sh

HOST=${0##*/}
HOST=${HOST#*-}
... 

The second part of the script

Code: Select all

umount /mnt/${HOST}/dev/shm
umount /mnt/${HOST}/dev
umount /mnt/${HOST}/proc
umount /mnt/${HOST}/sys
umount /mnt/${HOST}/usr/portage
umount /mnt/${HOST}/usr/local/portage
umount /mnt/${HOST}/var/tmp/portage
umount /mnt/${HOST}

is run after you have done upgrade/compiling isn't it?

[Joseph_sys]

I'm trying to combine information to make bash script to boot strap via NFS Box_slow on Box_fast to upgrade slower boxes.
I've found this link:
http://www.wikigentoo.ksiezyc.pl/Slow_systems.htm

and this script:

Code: Select all

#!/bin/sh

HOST=${0##*/}
HOST=${HOST#*-}

mkdir -p --mode=0755 /mnt/${HOST}

mount -t nfs -o rw,intr,noatime,actimeo=60,vers=4,fsc ${HOST}:/ /mnt/${HOST}
mount --bind /dev /mnt/${HOST}/dev
mount --bind /dev/shm /mnt/${HOST}/dev/shm
mount --bind /proc /mnt/${HOST}/proc
mount --bind /sys /mnt/${HOST}/sys
mount --bind /usr/portage /mnt/${HOST}/usr/portage
mount --bind /usr/local/portage /mnt/${HOST}/usr/local/portage
mount --bind /var/tmp/portage /mnt/${HOST}/var/tmp/portage

env -i - HOME="/root" TERM="$TERM" chroot /mnt/${HOST} /bin/bash -l

umount /mnt/${HOST}/dev/shm
umount /mnt/${HOST}/dev
umount /mnt/${HOST}/proc
umount /mnt/${HOST}/sys
umount /mnt/${HOST}/usr/portage
umount /mnt/${HOST}/usr/local/portage
umount /mnt/${HOST}/var/tmp/portage
umount /mnt/${HOST}

The part I is not clear to me in the above script is:

HOST=${0##*/}
HOST=${HOST#*-}

Shouldn't this bash script be in two parts:

Code: Select all

#!/bin/sh

HOST=${0##*/}
HOST=${HOST#*-}

mkdir -p --mode=0755 /mnt/${HOST}

mount -t nfs -o rw,intr,noatime,actimeo=60,vers=4,fsc ${HOST}:/ /mnt/${HOST}
mount --bind /dev /mnt/${HOST}/dev
mount --bind /dev/shm /mnt/${HOST}/dev/shm
mount --bind /proc /mnt/${HOST}/proc
mount --bind /sys /mnt/${HOST}/sys
mount --bind /usr/portage /mnt/${HOST}/usr/portage
mount --bind /usr/local/portage /mnt/${HOST}/usr/local/portage
mount --bind /var/tmp/portage /mnt/${HOST}/var/tmp/portage

env -i - HOME="/root" TERM="$TERM" chroot /mnt/${HOST} /bin/bash -l

Code: Select all

#!/bin/sh
umount /mnt/${HOST}/dev/shm
umount /mnt/${HOST}/dev
umount /mnt/${HOST}/proc
umount /mnt/${HOST}/sys
umount /mnt/${HOST}/usr/portage
umount /mnt/${HOST}/usr/local/portage
umount /mnt/${HOST}/var/tmp/portage
umount /mnt/${HOST}

To start NFS do I need start "portmap"?

Code: Select all

bash_server # rc-update add portmap default
bash_server # rc-update add nfs default
bash_client # rc-update add portmap default
bash_client # rc-update add nfsmount default
bash_client # /etc/init.d/nfsmount start 

[Chewi]

Can you explain me what these two line do in your script?

Code: Select all

#!/bin/sh

HOST=${0##*/}
HOST=${HOST#*-}
... 

If you create a symlink to this script like chroot-foo then HOST will be set to foo and it will try to mount the share from that server.

The second part of the script is run after you have done upgrade/compiling isn't it?

Yes but that script is old and there's a better way to do it now, which will only make the mounts visible inside the chroot and they will be more reliably unmounted when you exit.

Code: Select all

#!/bin/bash

HOST=${0##*/}
HOST=${HOST#*-}
ROOT=/mnt/${HOST}

mkdir -p --mode=0755 "${ROOT}"

exec sudo unshare -m /bin/bash -c "
set -e

mount -t nfs -o rw,noatime,nocto,actimeo=60,lookupcache=positive,vers=4,fsc '${HOST}:/' '${ROOT}'
mount --bind {,'${ROOT}'}/dev
mount --bind {,'${ROOT}'}/dev/pts
mount --bind {,'${ROOT}'}/dev/shm
mount --bind {,'${ROOT}'}/proc
mount --bind {,'${ROOT}'}/sys
mount --bind {,'${ROOT}'}/usr/local/portage
mount --bind {,'${ROOT}'}/usr/portage
mount --bind {,'${ROOT}'}/var/cache/edb/dep
mount --bind {,'${ROOT}'}/var/tmp/portage

exec chroot '${ROOT}' /bin/bash -l
"

[Joseph_sys]

If you create a symlink to this script like chroot-foo then HOST will be set to foo and it will try to mount the share from that server.

Thank you for quick reply.
I see, so I need to have a link file:
lrwxrwxrwx ......chroot-foo -> chroot.sh

In this case do I need to have a NFS entry in fstab called "foo" to point to the correct host?

The second part of the script is run after you have done upgrade/compiling isn't it?

Yes but that script is old and there's a better way to do it now, which will only make the mounts visible inside the chroot and they will be more reliably unmounted when you exit.

Code: Select all

#!/bin/sh

HOST=${0##*/}
HOST=${HOST#*-}
ROOT=/mnt/${HOST}

mkdir -p --mode=0755 "${ROOT}"

env -i - HOME="/root" TERM="${TERM}" exec sudo unshare -m /bin/sh -c "
set -e

mount -t nfs -o rw,noatime,nocto,actimeo=60,lookupcache=positive,vers=4,fsc '${HOST}:/' '${ROOT}'
mount --bind {,'${ROOT}'}/dev
mount --bind {,'${ROOT}'}/dev/pts
mount --bind {,'${ROOT}'}/dev/shm
mount --bind {,'${ROOT}'}/proc
mount --bind {,'${ROOT}'}/sys
mount --bind {,'${ROOT}'}/usr/local/portage
mount --bind {,'${ROOT}'}/usr/portage
mount --bind {,'${ROOT}'}/var/cache/edb/dep
mount --bind {,'${ROOT}'}/var/tmp/portage

exec chroot '${ROOT}' /bin/bash -l
"

Do you still need to have the second part of the script "umount" after you exit/close the terminal?

I run onto this wiki link:
http://www.wikigentoo.ksiezyc.pl/Slow_systems.htm

Code: Select all

# mount <IP of box B>:/ /mnt/slowmachine -o rsize=1024,wsize=1024,rw

(The -o rsize=1024,wsize=1024 prevents IP fragmentation, rw permits read-write operation on the NFS mount.)

Is the above entry useful?

Also in the part NFS mount; Is starting "portmap" necessary?
http://www.wikigentoo.ksiezyc.pl/Shared ... ia_NFS.htm

Code: Select all

bash_server # rc-update add portmap default
bash_server # rc-update add nfs default

bash_client # rc-update add portmap default
bash_client # rc-update add nfsmount default
bash_client # /etc/init.d/nfsmount start 

[Joseph_sys]

I've tried to duplicate your script but got:

Code: Select all

sh chroot-10.10.0.2 
env: ‘exec’: No such file or directory

10.10.0.2 is my local IP box running "NFS"

Code: Select all

ll chroot*
lrwxrwxrwx 1 root root   9 Mar 28 18:31 chroot-10.10.0.2 -> chroot.sh
-rwxr--r-- 1 root root 634 Mar 28 18:32 chroot.sh

Code: Select all

cat chroot.sh 
#!/bin/sh

HOST=${0##*/}
HOST=${HOST#*-}
ROOT=/mnt/${HOST}

mkdir -p --mode=0755 "${ROOT}"

env -i - HOME="/root" TERM="${TERM}" exec sudo unshare -m /bin/sh -c "
set -e

mount -t nfs -o rw,noatime,nocto,actimeo=60,lookupcache=positive,vers=4,fsc '${HOST}:/' '${ROOT}'
mount --bind {,'${ROOT}'}/dev
mount --bind {,'${ROOT}'}/dev/pts
mount --bind {,'${ROOT}'}/dev/shm
mount --bind {,'${ROOT}'}/proc
mount --bind {,'${ROOT}'}/sys
mount --bind {,'${ROOT}'}/usr/local/portage
mount --bind {,'${ROOT}'}/usr/portage
mount --bind {,'${ROOT}'}/var/cache/edb/dep
mount --bind {,'${ROOT}'}/var/tmp/portage

exec chroot '${ROOT}' /bin/bash -l
"

[krinn]

The part I is not clear to me in the above script is:
HOST=${0##*/}
HOST=${HOST#*-}

It's a little trick, it extract the name of the execute script and kick out the path from it (1st line)
2nd line, clear what is in front of -
Why doing that? because you can setup symlink with the hostname in it and it will find the hostname to use from the symlink
ie: ln -s thisscript thisscript-somehost
result in :
* /usr/bin/thiscript-somehost -> thiscript-somethost
* thisscript-somehost -> somehost

Shouldn't this bash script be in two parts:

Not really, the disturbing part is why mount --bind all these directories when you just unmount them just after?
Because in between, the script enter inside a chroot (the env -i - HOM...) and so it's only when you exit that chroot that the script continue with the umount commands

To start NFS do I need start "portmap"?

No, the script create nfsv4 share, no need for portmap.
Anyway, when you want nfs openrc would start needed dep if need

[Joseph_sys]

The part I is not clear to me in the above script is:
HOST=${0##*/}
HOST=${HOST#*-}

It's a little trick, it extract the name of the execute script and kick out the path from it (1st line)
2nd line, clear what is in front of -
Why doing that? because you can setup symlink with the hostname in it and it will find the hostname to use from the symlink
ie: ln -s thisscript thisscript-somehost
result in :
* /usr/bin/thiscript-somehost -> thiscript-somethost
* thisscript-somehost -> somehost

Shouldn't this bash script be in two parts:

Not really, the disturbing part is why mount --bind all these directories when you just unmount them just after?
Because in between, the script enter inside a chroot (the env -i - HOM...) and so it's only when you exit that chroot that the script continue with the umount commands

To start NFS do I need start "portmap"?

No, the script create nfsv4 share, no need for portmap.
Anyway, when you want nfs openrc would start needed dep if need

Thank you for replying. I have an updated form of script.

Code: Select all

#!/bin/sh

HOST=${0##*/}
HOST=${HOST#*-}
ROOT=/mnt/${HOST}

mkdir -p --mode=0755 "${ROOT}"

env -i - HOME="/root" TERM="${TERM}" exec sudo unshare -m /bin/sh -c "
set -e

mount -t nfs -o rw,noatime,nocto,actimeo=60,lookupcache=positive,vers=4,fsc '${HOST}:/' '${ROOT}'
mount --bind {,'${ROOT}'}/dev
mount --bind {,'${ROOT}'}/dev/pts
mount --bind {,'${ROOT}'}/dev/shm
mount --bind {,'${ROOT}'}/proc
mount --bind {,'${ROOT}'}/sys
mount --bind {,'${ROOT}'}/usr/local/portage
mount --bind {,'${ROOT}'}/usr/portage
mount --bind {,'${ROOT}'}/var/cache/edb/dep
mount --bind {,'${ROOT}'}/var/tmp/portage

exec chroot '${ROOT}' /bin/bash -l
"

So I created:

Code: Select all

lrwxrwxrwx 1 root root   9 Mar 28 18:31 chroot-10.10.0.2 -> chroot.sh
-rwxr--r-- 1 root root 634 Mar 28 18:32 chroot.sh

10.10.0.2 is my local IP box running "NFS"

When I run: "sh chroot-10.10.0.2" I get:

Code: Select all

sh chroot-10.10.0.2
env: ‘exec’: No such file or directory

I think I'm doing something wrong.

On a NFS-server I have:

Code: Select all

cat  /etc/exports
# /etc/exports: NFS file systems being exported.  See exports(5).
/ 10.10.0.5(rw,no_root_squash,sync,no_subtree_check)

[Hu]

Do you still need to have the second part of the script "umount" after you exit/close the terminal?

No. Chewi switched to using a private mount namespace specifically to resolve that problem. Since the mount commands run in an unshared mount namespace, they have no impact on processes outside the namespace. When the last process in the namespace exits, the namespace ceases to exist and all mounts in it are automatically undone by the kernel.

[krinn]

Have you checked which process is blocked during this time? Specifically, is it stalled waiting for a mount to finish or is it stalled waiting for the chroot'd bash to initialize?

For me it's bash, because he is mounting the "host" /proc, /sys... inside the chroot, and they are invalid for the build host and refers to nothing.
ie: /proc/118 is a valid pid for the "host", but invalid for the build host, as 118 pid may not exists at all, and this goes for all files and entries

It's also dangerous, because the "host" values may clash with the "build host" and something nasty could happen!
ie: /dev/disk/by-id/totally_safe_to_clean point to /dev/sdb1 while /dev/sdb1 in the build host is not really something you can safely clean

While i get the idea why you do that Chewi, i have a bad feeling about it
I would myself only mount the "host" /proc, /sys... only if an ebuild fails because it is seeking infos

[pjp]

I'm trying to combine information to make bash script to boot strap via NFS Box_slow on Box_fast to upgrade slower boxes.
I've found

Not ideal, but merged this thread and its couple of posts as there was too much crossover.

[Chewi]

Apologies, I didn't test this script as much as I should have because I'd only given the "unshare" treatment to my local chroot script. I now remember that the env stuff isn't really necessary because sudo will effectively do that anyway. Change that line to:

Code: Select all

exec sudo unshare -m /bin/sh -c "

You don't need to have an entry in fstab but you could adjust the script to work that way if you prefer. I see rsize and wsize options used a lot but the man page says that the largest supported value will always be used by default and I have found that to be the case with NFSv4.

For me it's bash, because he is mounting the "host" /proc, /sys... inside the chroot, and they are invalid for the build host and refers to nothing.
ie: /proc/118 is a valid pid for the "host", but invalid for the build host, as 118 pid may not exists at all, and this goes for all files and entries

It's also dangerous, because the "host" values may clash with the "build host" and something nasty could happen!
ie: /dev/disk/by-id/totally_safe_to_clean point to /dev/sdb1 while /dev/sdb1 in the build host is not really something you can safely clean

While i get the idea why you do that Chewi, i have a bad feeling about it
I would myself only mount the "host" /proc, /sys... only if an ebuild fails because it is seeking infos

I think you're confused. I have called unshare with -m, which only unshares the mount namespace. The PID namespace remains shared, just as it always has been when chrooting in the traditional manner. I've checked the Gentoo handbook and, true enough, it does tell you to mount a new /proc instance instead of bind mounting it. I'm not sure whether it said that back in 2002 when I first installed Gentoo but as far as I can tell, it doesn't make any difference unless you unshare the PID namespace as well. If I start a chroot and run ps, I can see all the processes from outside the chroot. Conversely, if I start nano in my chroot and run ps outside of it, I can see that nano instance in the listing and under /proc. Of course, you can unshare the other namespaces if you want to, which leads into things like LXC and Docker but I've never played with that. As for /dev, I don't think that is ever unshared. /dev/sda is /dev/sda regardless of which chroot you're in and things like /dev/disk/by-id are handled by the kernel.

[krinn]

I think you're confused.

Actually yes, but not by what you think, i mistake bind arguments for "dir dev" instead of "dev dir"

[Joseph_sys]

Apologies, I didn't test this script as much as I should have because I'd only given the "unshare" treatment to my local chroot script. I now remember that the env stuff isn't really necessary because sudo will effectively do that anyway. Change that line to:

Code: Select all

exec sudo unshare -m /bin/sh -c "

[snip]

I think it worked. THANK YOU!
I've added "set -x" to the beginning of the script and run a link file:
"sh chroot-eden" --> chroot.sh

Code: Select all

+ HOST=chroot-eden
+ HOST=eden
+ ROOT=/mnt/eden
+ mkdir -p --mode=0755 /mnt/eden
+ exec sudo unshare -m /bin/sh -c '
set -e

mount -t nfs -o rw,noatime,nocto,actimeo=60,lookupcache=positive,vers=4,fsc '\''eden:/'\'' '\''/mnt/eden'\''
mount --bind {,'\''/mnt/eden'\''}/dev
mount --bind {,'\''/mnt/eden'\''}/dev/pts
mount --bind {,'\''/mnt/eden'\''}/dev/shm
mount --bind {,'\''/mnt/eden'\''}/proc
mount --bind {,'\''/mnt/eden'\''}/sys
mount --bind {,'\''/mnt/eden'\''}/usr/local/portage
mount --bind {,'\''/mnt/eden'\''}/usr/portage
mount --bind {,'\''/mnt/eden'\''}/var/cache/edb/dep
mount --bind {,'\''/mnt/eden'\''}/var/tmp/portage

exec chroot '\''/mnt/eden'\'' /bin/bash -l
'

eden - small (old) box running nfs server
syscon3 - my 8-core AMD

I was presented with root prompt "syscon3 /" and I know I'm in "chroot" as I created a file in / 1.txt and it appeared when I typed on "ls -al"
However, when I logged from another shell to "syscon3" box the /mnt/eden/ dir is empty. Where are the files mounted?

How do I change the prompt to be: "syscon3-eden"?
It would be less confusing and an indication that I'm in chroot environment.

[Chewi]

Great!

I was presented with root prompt "syscon3 /" and I know I'm in "chroot" as I created a file in / 1.txt and it appeared when I typed on "ls -al"
However, when I logged from another shell to "syscon3" box the /mnt/eden/ dir is empty. Where are the files mounted?

Well that's the thing with unshared mount namespaces, you won't see them from another shell. If you just want another session, you can call the script again, or you could use tmux or similar. If you want to move files around then you could mount the NFS share separately in fstab and just leave the script to do the other mounts. If it's not available all the time, there is autofs but I've not tried that in years.

How do I change the prompt to be: "syscon3-eden"?
It would be less confusing and an indication that I'm in chroot environment.

I've never tried it but you could unshare the UTS namespace as well, allowing an independent host and domain name. I guess you'd need to set it each time as part of the script. You could also manipulate the Bash prompt (PS1) based on some environment variable.

[Joseph_sys]

Great!

I was presented with root prompt "syscon3 /" and I know I'm in "chroot" as I created a file in / 1.txt and it appeared when I typed on "ls -al"
However, when I logged from another shell to "syscon3" box the /mnt/eden/ dir is empty. Where are the files mounted?

Well that's the thing with unshared mount namespaces, you won't see them from another shell. If you just want another session, you can call the script again, or you could use tmux or similar. If you want to move files around then you could mount the NFS share separately in fstab and just leave the script to do the other mounts. If it's not available all the time, there is autofs but I've not tried that in years.

How do I change the prompt to be: "syscon3-eden"?
It would be less confusing and an indication that I'm in chroot environment.

I've never tried it but you could unshare the UTS namespace as well, allowing an independent host and domain name. I guess you'd need to set it each time as part of the script. You could also manipulate the Bash prompt (PS1) based on some environment variable.

This is a live test. My old system has not been upgraded for about 250-days. So I emerge --sync. and try to run in chroot evironment:

Code: Select all

emerge --ask --oneshot -vq sys-devel/gcc 
...
libtool: install: warning: remember to run `libtool --finish /usr/lib/../lib'
make[4]: Nothing to be done for 'install-data-am'.
make[4]: Leaving directory '/var/tmp/portage/sys-devel/gcc-6.4.0-r1/work/build/i686-pc-linux-gnu/libatomic'
make[3]: Leaving directory '/var/tmp/portage/sys-devel/gcc-6.4.0-r1/work/build/i686-pc-linux-gnu/libatomic'
make[2]: Leaving directory '/var/tmp/portage/sys-devel/gcc-6.4.0-r1/work/build/i686-pc-linux-gnu/libatomic'
make[1]: Leaving directory '/var/tmp/portage/sys-devel/gcc-6.4.0-r1/work/build'
 * PT_PAX marking -r /var/tmp/portage/sys-devel/gcc-6.4.0-r1/image//usr/libexec/gcc/i686-pc-linux-gnu/6.4.0/cc1 with scanelf
 * XATTR_PAX marking -re /var/tmp/portage/sys-devel/gcc-6.4.0-r1/image//usr/libexec/gcc/i686-pc-linux-gnu/6.4.0/cc1 with setfattr
 * PT_PAX marking -r /var/tmp/portage/sys-devel/gcc-6.4.0-r1/image//usr/libexec/gcc/i686-pc-linux-gnu/6.4.0/cc1plus with scanelf
 * XATTR_PAX marking -re /var/tmp/portage/sys-devel/gcc-6.4.0-r1/image//usr/libexec/gcc/i686-pc-linux-gnu/6.4.0/cc1plus with setfattr
 * Final size of build directory: 1556216 KiB
 * Final size of installed tree: 151528 KiB
!!! Failed to copy extended attributes. In order to avoid this error,
!!! set FEATURES="-xattr" in make.conf.
!!! copy /var/tmp/portage/sys-devel/gcc-6.4.0-r1/image/usr/libexec/gcc/i686-pc-linux-gnu/6.4.0/cc1 -> /usr/libexec/gcc/i686-pc-linux-gnu/6.4.0/cc1 failed.
!!! Filesystem containing file '/usr/libexec/gcc/i686-pc-linux-gnu/6.4.0/cc1#new' does not support extended attribute 'user.pax.flags'
 * 
 * Please include /var/tmp/portage/sys-devel/gcc-6.4.0-r1/work/gcc-build-logs.tar.bz2 in your bug report.
 * 
 * 
 * The following package has failed to build, install, or execute postinst:
 * 
 *  (sys-devel/gcc-6.4.0-r1:6.4.0/6.4.0::gentoo, ebuild scheduled for merge), Log file:
 *   '/var/tmp/portage/sys-devel/gcc-6.4.0-r1/temp/build.log'
 * 

Code: Select all

Portage 2.3.6 (python 3.4.5-final-0, default/linux/x86/13.0/desktop, gcc-4.9.4, glibc-2.23-r4, 4.9.72-gentoo x86_64)
=================================================================
System uname: Linux-4.9.72-gentoo-x86_64-AMD_Ryzen_5_1400_Quad-Core_Processor-with-gentoo-2.3
KiB Mem:    16432224 total,  12064148 free
KiB Swap:     524284 total,    524284 free
Timestamp of repository gentoo: Mon, 26 Mar 2018 21:00:01 +0000
sh bash 4.3_p48-r1
ld GNU ld (Gentoo 2.28 p1.2) 2.28
app-shells/bash:          4.3_p48-r1::gentoo
dev-java/java-config:     2.2.0-r3::gentoo
dev-lang/perl:            5.24.1-r2::gentoo
dev-lang/python:          2.7.12::gentoo, 3.4.5::gentoo
dev-util/cmake:           3.7.2::gentoo
dev-util/pkgconfig:       0.28-r2::gentoo
sys-apps/baselayout:      2.3::gentoo
sys-apps/openrc:          0.26.3::gentoo
sys-apps/sandbox:         2.10-r3::gentoo
sys-devel/autoconf:       2.13::gentoo, 2.69::gentoo
sys-devel/automake:       1.15-r2::gentoo
sys-devel/binutils:       2.28-r2::gentoo
sys-devel/gcc:            4.9.4::gentoo, 5.4.0-r3::gentoo
sys-devel/gcc-config:     1.7.3::gentoo
sys-devel/libtool:        2.4.6-r3::gentoo
sys-devel/make:           4.2.1::gentoo
sys-kernel/linux-headers: 4.4::gentoo (virtual/os-headers)
sys-libs/glibc:           2.23-r4::gentoo
Repositories:

gentoo
    location: /usr/portage
    sync-type: rsync
    sync-uri: rsync://10.0.0.103/gentoo-portage
    priority: -1000

x-portage
    location: /usr/local/portage
    masters: gentoo
    priority: 0

brother-overlay
    location: /var/lib/layman/brother-overlay
    masters: gentoo
    priority: 50

ACCEPT_KEYWORDS="x86"
ACCEPT_LICENSE="* -@EULA PUEL dlj-1.1"
CBUILD="i686-pc-linux-gnu"
CFLAGS="-O2 -march=i686 -pipe"
CHOST="i686-pc-linux-gnu"
CONFIG_PROTECT="/etc /usr/lib/fax /usr/share/easy-rsa /usr/share/gnupg/qualified.txt /usr/src/linux* /var/spool/fax/etc"
CONFIG_PROTECT_MASK="/etc/ca-certificates.conf /etc/dconf /etc/env.d /etc/fonts/fonts.conf /etc/gconf /etc/gentoo-release /etc/revdep-rebuild /etc/sandbox.d /etc/terminfo /etc/texmf/language.dat.d /etc/texmf/language.def.d /etc/texmf/updmap.d /etc/texmf/web2c"
CXXFLAGS="-O2 -march=i686 -pipe"
DISTDIR="/usr/portage/distfiles"
EMERGE_DEFAULT_OPTS="--autounmask-write=y --keep-going --with-bdeps=y"
FCFLAGS="-O2 -march=i686 -pipe"
FEATURES="assume-digests binpkg-logs config-protect-if-modified distlocks ebuild-locks fixlafiles merge-sync news parallel-fetch preserve-libs protect-owned sandbox sfperms strict unknown-features-warn unmerge-logs unmerge-orphans userfetch userpriv usersandbox usersync xattr"
FFLAGS="-O2 -march=i686 -pipe"
GENTOO_MIRRORS="http://gentoo.llarian.net/ http://gentoo.mirrors.hoobly.com/ http://ftp.ucsb.edu/pub/mirrors/linux/gentoo/ http://gentoo.mirrors.easynews.com/linux/gentoo/"
LANG="en_US.UTF-8"
LC_ALL="en_US.UTF-8"
LDFLAGS="-Wl,-O1 -Wl,--as-needed"
MAKEOPTS="-j2"
PKGDIR="/usr/portage/packages"
PORTAGE_CONFIGROOT="/"
PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --omit-dir-times --compress --force --whole-file --delete --stats --human-readable --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages --exclude=/.git"
PORTAGE_TMPDIR="/var/tmp"
USE="X a52 aac acpi alsa apache2 bluetooth branding bzip2 cairo cdda cdr cgi cli consolekit crypt cups cxx dbus dri dts dvd dvdr emboss encode exif fam flac fortran gdbm gif glamor gpm gtk iconv ipv6 jpeg lcms ldap libnotify mad mng modules mp3 mp4 mpeg ncurses nls nptl ogg opengl openmp pam pango pcre pdf png policykit ppds qt3support qt5 readline scanner sdl seccomp spell ssl startup-notification svg tcpd tiff truetype type1 udev udisks unicode upower usb vorbis wxwidgets x264 x86 xattr xcb xml xv xvid zlib" ABI_X86="32" ALSA_CARDS="ali5451 als4000 atiixp atiixp-modem bt87x ca0106 cmipci emu10k1 emu10k1x ens1370 ens1371 es1938 es1968 fm801 hda-intel intel8x0 intel8x0m maestro3 trident usb-audio via82xx via82xx-modem ymfpci" APACHE2_MODULES="authn_core authz_core socache_shmcb unixd actions alias auth_basic authn_alias authn_anon authn_dbm authn_default authn_file authz_dbm authz_default authz_groupfile authz_host authz_owner authz_user autoindex cache cgi cgid dav dav_fs dav_lock deflate dir disk_cache env expires ext_filter file_cache filter headers include info log_config logio mem_cache mime mime_magic negotiation rewrite setenvif speling status unique_id userdir usertrack vhost_alias" CALLIGRA_FEATURES="kexi words flow plan sheets stage tables krita karbon braindump author" COLLECTD_PLUGINS="df interface irq load memory rrdtool swap syslog" ELIBC="glibc" GPSD_PROTOCOLS="ashtech aivdm earthmate evermore fv18 garmin garmintxt gpsclock isync itrax mtk3301 nmea ntrip navcom oceanserver oldstyle oncore rtcm104v2 rtcm104v3 sirf skytraq superstar2 timing tsip tripmate tnt ublox ubx" INPUT_DEVICES="evdev" KERNEL="linux" L10N="en" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" LIBREOFFICE_EXTENSIONS="presenter-console presenter-minimizer" OFFICE_IMPLEMENTATION="libreoffice" PHP_TARGETS="php5-6 php7-0" POSTGRES_TARGETS="postgres9_5" PYTHON_SINGLE_TARGET="python3_5" PYTHON_TARGETS="python2_7 python3_5" RUBY_TARGETS="ruby22 ruby23" SANE_BACKENDS="fujitsu" USERLAND="GNU" VIDEO_CARDS="vga vesa fbdev via" XTABLES_ADDONS="quota2 psd pknock lscan length2 ipv4options ipset ipp2p iface geoip fuzzy condition tee tarpit sysrq steal rawnat logmark ipmark dhcpmac delude chaos account"
Unset:  CC, CPPFLAGS, CTARGET, CXX, INSTALL_MASK, LINGUAS, PORTAGE_BUNZIP2_COMMAND, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS

[Chewi]

So do what it says. NFS doesn't support extended attributes. This is not a hardened system so you could also set PAX_MARKINGS="none" in make.conf to stop it trying to apply them.

[Joseph_sys]

So do what it says. NFS doesn't support extended attributes. This is not a hardened system so you could also set PAX_MARKINGS="none" in make.conf to stop it trying to apply them.

That scrip worked very well! THANK YOU!

I've "dusted-off" and upgraded my old system:
VIA Eden Processor 1200MHz 1GB or RAM (only)

The system wasn't upgraded in over 250-days so I upgraded to gcc-6.4.0-r1 and recompile 756-packages (-e @world) running on
8-core AMD with 16GB or RAM

This scrip should be in the official Gentoo Documentation, may old systems are getting discarded as there are two small to compile newer/and larger packages.

The only think that puzzle me I don't get the the prompt "PS1=eden"

When I run:

Code: Select all

syscon3 /home/joseph # sh chroot-eden 
+ HOST=chroot-eden
+ HOST=eden
+ ROOT=/mnt/eden
+ PS1=eden
+ mkdir -p --mode=0755 /mnt/eden
+ exec sudo unshare -m /bin/sh -c '
set -e

mount -t nfs -o rw,noatime,nocto,actimeo=60,lookupcache=positive,vers=4,fsc '\''eden:/'\'' '\''/mnt/eden'\''
mount --bind {,'\''/mnt/eden'\''}/dev
mount --bind {,'\''/mnt/eden'\''}/dev/pts
mount --bind {,'\''/mnt/eden'\''}/dev/shm
mount --bind {,'\''/mnt/eden'\''}/proc
mount --bind {,'\''/mnt/eden'\''}/sys
mount --bind {,'\''/mnt/eden'\''}/usr/local/portage
mount --bind {,'\''/mnt/eden'\''}/usr/portage
mount --bind {,'\''/mnt/eden'\''}/var/cache/edb/dep
mount --bind {,'\''/mnt/eden'\''}/var/tmp/portage

exec chroot '\''/mnt/eden'\'' /bin/bash -i
'
syscon3 / #

The "PS1=eden" when when the prompt shows up, I end-up with "syscon3 / #"

All the 756-packages recompile in about 24-hours.
gcc-6.4.0-r1 took only 1hr 39min

The packages that wouldn't compile is binary:
virtualbox-bin