NikosAlexandris
-
Posts
111 -
Joined
-
Last visited
-
Days Won
1
Content Type
Profiles
Forums
Blogs
Posts posted by NikosAlexandris
-
-
In each power-off process, be it `shutdown` or `reboot`, there is this failure that's printed just before powering actually off:
/lib64/rc/cache not writable!
...I read through the (few) related threads on the web for it, like in https://archives.gentoo.org/gentoo-user/message/3e236160b893faf12eefed55283a8970.
As well, I think it is the same issue discussed (search for /lib64/rc) at http://www.mattzone.com/zfsonlinux/zfsonlinux_20140130.html
-
anything like /var or /usr used as separate zfs mounts?
If zfs mounts are meant to be the mount points of ZFS datasets, then at least the one from the installation guide:
# zfs create -o mountpoint=/var/tmp/portage -o compression=lz4 -o sync=disabled rpool/FUNTOO/build
In my very case, at the moment (while I plan to further split various system directories to occupy their own ZFS dataset):
... xpool/FUNTOO/build 1.73G 302G 1.73G /var/tmp/portage ...
-
I too have noticed this when testing Funtoo on ZFS root. The storage pool containing root is exported during shutdown so this message suggests that the export is occurring before OpenRC is finished with writing to /lib64/rc/cache.
The seriousness of this is debatable given that the system restarts without any obvious problems. Some research would be required.
I remember reading this message for the last 2 years, always on ZFS, in laptop and desktop workstation. The only problems I have had were right in the beginning, the first reboot(s), after completing the installation of Funtoo on ZFS and when building a new kernel. Albeit, it is known that re-building another (or newer) kernel needs to remember to re-compile ZFS related modules. Since I rebooted quite some times the last days, while testing various kernel configurations, I decided it's time to deal with it.
(Off topic?) However, I observe random freezing while booting the laptop the last days -- the boot sequence freezes, some times, at random points. I am on
uname -a Linux carbon 4.5.0-pf4 #1 SMP Mon Nov 21 11:36:43 CET 2016 x86_64 Intel(R) Core(TM) i7-4600U CPU @ 2.10GHz GenuineIntel GNU/Linux
Maybe the cause is some kernel mis-configuration (as I strive to customize the kernel to just fit to the hardware I run). I tend to believe it happens when booting without AC plugged-in.
-
I hope this is the right forum for the following question. I happily run ZFS on Funtoo. In each power-off process, be it `shutdown` or `reboot`, there is this failure that's printed just before powering actually off: `/lib64/rc/cache not writable!`.
The directory exists and belongs to root:
stat /lib64/rc/cache/ File: ?/lib64/rc/cache/? Size: 5 Blocks: 1 IO Block: 512 directory Device: 12h/18d Inode: 481694 Links: 2 Access: (0755/drwxr-xr-x) Uid: ( 0/ root) Gid: ( 0/ root) Access: 2016-11-22 03:23:11.605391160 +0100 Modify: 2016-10-24 23:20:18.612038472 +0200 Change: 2016-11-22 04:54:57.727073038 +0100 Birth: -
Is this important? How to proceed in understanding the problem? I read through the (few) related threads on the web for it, like in https://archives.gentoo.org/gentoo-user/message/3e236160b893faf12eefed55283a8970. No solution(s) however offered.
-
Thank you for the latest update of the ZFS Install Guide.
I still cannot boot into it successfully. Please, have a look at some points I left at http://www.funtoo.org/Talk:ZFS_Install_Guide.
-
Yes, I read this. But, this can be worked-around. And, in fact, it's not the problem. Installing GRUB is the problem. Is there a way to first go through the normal installtion process and have working GRUB. Then, migrate almost everything to ZFS?
-
I got past the problem with zfs. I can see the zpool status report. The only missing part is to install grub. I don't understand why `grub-install` does not find the EFI directory. There is a valid partition marked as EF02. I guess I have to report this with more details.
-
Trying to follow up the installation process as described in <http://www.funtoo.org/ZFS_Install_Guide>.
Only yesterday, however, I discovered the tutorial at <http://www.funtoo.org/Install_ZFS_root%26boot_File_System>. I think I understand, now, after reading the notes there-in the problem I am stuck on (read below). But don't know how to work-it-around. I use(d) sysresccd-4.7.0_zfs_0.6.5.4. After staging and installing portage (both funtoo-stable), I am to the point of confirming that the zfs pool can be read.
I get the following error:
```
The device /dev/zfs is missing and must be created.
Try running 'udevadm trigger' as root to create it.
```
The `/dev/zfs` exists however. And outside of the "chroot", the pool and the datasets are correctly reported.
I guess this all boils down to "5. Upgrading to zfs version >= 0.6.5.4" as per the notes in <http://www.funtoo.org/Install_ZFS_root%26boot_File_System>.I did install zfs and zfs-kmod version 0.6.5.4-r1 (this is the one installed when requesting for emerge sys-zfs/zfs-0.6.5.4 at the moment). And still, the zpool utility does not work.
Since the initramfs/kernel in sysresccd-4.7.0_zfs_0.6.5.4 is pre zfs-0.6.5.3 and does not understand zfs-0.6.5.4 (!?), how does an update of the kernel and initramfs, using genkernel, e.g.
```
# genkernel all --zfs --no-clean --kerneldir=/usr/src/linux -- kernel-config=/usr/src/<path_to_config> --callback="emerge -1 spl zfs-kmod zfs"
```
will solve the issue, if I can't "boot this kernel"?I opted for having a dedicated boot partition (/dev/sda1) and not installing everything on ZFS.
Still, I can't make any use of grub, i.e. install it in /dev/sda. The command `grub-install` does respond with
```
Installing for x86_64-efi platform.
grub-install: error: cannot find EFI directory.
```
Any pointers?
-
Too complicated and time consuming, if not dangerous. The `emerge -avp @downgrade` shows much too many problems. I'll stick with current. I have to find some time in the near future to re-install. The main reason is that I want really a more stable system, less updates.
Thank you for the insights!
-
Thank you Sandro.
..
re-configure your build with "# epro build stable"
take a look in /etc/portage/make.conf if are enabled ~amd64 (then delete ~amd64"
..
Just a clarification required -- why/how is `~amd64` relevant to the process?
-
Is it a bad idea to switch the build type from `current` to `stable`? Using `epro build stable`?
There is a warning (reported in the command line, as well as in this "epro" wiki-page: http://www.funtoo.org/Funtoo_Profiles):epro build stable WARNING: build already set to current -- typically, user should not change this.
-
Thanks for this one. I needed it too!
-
What is the best way to make `keychain` and `kwallet` co-exist? It it possible to just give a passphrase once (given, `keychain` is set up correctly, see funtoo.org/Keychain) for everything (and, re-enter after the `keychain`'s for example `--timeout 180` parameter has expired)? I need keychain for mutt and ssh-ing. Can't it be all right at the login?
Some references (but I am still not sure how to proceed):
-
Besides the binary (under /usr/sbin), the include files, the librairies and various xml files (under /usr/share), there is only one related conf file:
/etc/dbus-1/system.d/org.freedesktop.ModemManager1.conf
There is no related ModemManager entry in /etc/init.d. I will try to create a "service" out of it. If this won't work out, then I will try the "local.d" suggestion. Note, the Mobile Broadband option appears automatically once ModemManager is launched manually (as root, of course), while NetworkManager is doing its work, and it works as well. Thus, there is probably no need to restart the NetworkManager.
Thank you for your time.
-
I am manually connecting to WWAN via a Sierra EM7345 modem. The NetworkManager doens't seem to start ModemManager. Only if the latter is ran manually, it then appeas in NM. It is the same problem discussed at http://forums-web2.gentoo.org/viewtopic-t-994224.html?sid=643d116182decb2cf23a9d0d6a43ca9a.
Is there a way to fix this? Add MM as a service (openrc)?
-
For some reason, Kgpg, after the updates, did not keep to play nice along with keychain. Me thinks, at least. The gpg-agent was not running actually. I don't know how this relates to ssh-add not being executed with the keychain call. Anyhow, I unmerged Kgpg, and added
- the use-agent instruction in ~/.gnupg/gpg.conf
- the --agents ssh,gpg instruction to the keychain call in ~/.bash_profile in order to ensure both agents are launched. It seems that this fixed my problem.
Any best practice hints to work with both, keychain and kgpg? Or don't mix-em up?
-
Thanks. After trial and error, I think the problem, at least for ssh and scp, is that ssh-add is not persistent between reboots. It was not the case one week ago, prior my latest world updates.
More info on the directory's permissions
700 drwx------ 2 nik 9 Jul 26 15:52 .ssh
and of some of its content
700 drwx------ 2 nik 9 Jul 26 15:52 . 755 drwxr-xr-x 56 nik 213 Aug 9 12:30 .. 600 -rw------- 1 nik 739 May 22 15:24 authorized_keys 644 -rw-r--r-- 1 nik 122 Jul 11 18:33 config 600 -rw------- 1 nik 1.8K May 22 15:03 id_rsa 644 -rw-r--r-- 1 nik 404 May 22 15:03 id_rsa.pub 644 -rw-r--r-- 1 nik 1.4K Aug 4 20:59 known_hosts
I read here that the config file's permissions, should be set to 600. Which I have not. And, likewise, the ones of authorized_keys. The directory's should be 700. Do we have any reference for ssh related file permissions?
-
I have set up keychain to use an id_rsa key. After latest world updates, I am asked for a passphrase repeatedly. I got the following files/entries:
~/.ssh:
authorized_keys
config
id_rsaid_rsa.pub
known_hosts..
/etc/ssh/sshd_config (perhaps irrelevant here, however, it is the same setup used in other system into which I ssh from the current one)
PermitRootLogin no PubkeyAuthentication yes PasswordAuthentication no PermitEmptyPasswords no UsePAM no AllowTcpForwarding yes X11Forwarding yes AcceptEnv LANG LC_*
~/.gnupg:
gpg-agent.conf
no-grab default-cache-ttl 28800z default-cache-ttl-ssh 28800z max-cache-ttl 28800z max-cache-ttl-ssh 28800z
gpg.conf
# GnuPG config file created by KGpg use-agent default-key 9053534B693C4FB3 encrypt-to 9053534B693C4FB3
private-keys-v1.d (directory)
pubring.gpg
random_seed
secring.gpg
trustdb.gpg..
~/.keychain/
..tpx1c2g-sh
tpx1c2g-sh-gpg~/bash_profile
# GPG #export GPG_TTY=`tty` export GNUPGHOME=~/.gnupg export GPGKEY=693C4FB3 eval `keychain --noask --eval --timeout 180 id_rsa` source ~/.keychain/$HOSTNAME-sh source ~/.keychain/$HOSTNAME-sh-gpg
I have added the Key, again, via ssh-add to be sure it is used. I am still not sure why it stopped working, nor why it is not working now, even after checking that ssh-agent is (already) running. What else should I check?
-
Closing this topic (though not exactly solved). I re-installed Funtoo :-)
-
bliss-boot didn't work again for me. boot-update works fine. Though, the initrd and vmlinuz files have to be under /boot to be detected. Took me a while to figure out as I was moving the initrd related file inside /boot/kernels/Some-Linux-Kernel-Version, as required by bliss-boot.
-
Hi Peje,
if you have a look in the posts of mine (above), I do use as well bliss-initramfs (i.e. bliss-initramfs 1 3.12.22-KS.01). My boot.conf file is posted as well. If you spot anything wrong into it, could you please highlight it for me? Thank you.
-
I can't say it's easy. I've been trying the suggested flags but I get blockers and other problems as well. I decided to go the hard way. Meaning, I remove anything that blocks and mask anything that isn't installed but still seems to block the update process. So far, I've removed many packages which, however, are related to kde applications and other tools that are not directly bound to the system (me thinks at least).
Currently, I have a problem with cblas-reference which fails during the compile phase. Will post details (from the machine to-be-updated as soon as time permits).
-
I am trying to upgrade a Funtoo - workstation (completely irrelevant with http://forums.funtoo.org/index.php?/topic/82-zfs-pool-not-imported-at-boot-home-inaccessible/),
uname -a Linux Resilience 3.2.41-2 #1 SMP Tue Dec 10 21:50:32 EET 2013 x86_64 Intel(R) Xeon(R) CPU E5-2630 0 @ 2.30GHz GenuineIntel GNU/Linux
via
emerge -uavDN @world
Despite numerous attempts (providing suggested USE flags, masking packages or, even, removing some of the packages that appear to block the process), I always get something like the following as a response:
These are the packages that would be merged, in order: Calculating dependencies... done! !!! The ebuild selected to satisfy ">=net-libs/webkit-gtk-1.6:3=" has unmet requirements. - net-libs/webkit-gtk-2.4.3::gentoo USE="X egl geoloc gstreamer jit opengl webgl (-aqua) -coverage -debug -gles2 -introspection -libsecret -spell -test -wayland" The following REQUIRED_USE flag constraints are unsatisfied: geoloc? ( introspection ) The above constraints are a subset of the following complete expression: geoloc? ( introspection ) introspection? ( gstreamer ) gles2? ( egl ) webgl? ( exactly-one-of ( gles2 opengl ) ) !webgl? ( at-most-one-of ( gles2 opengl ) ) any-of ( aqua wayland X ) (dependency required by "net-libs/libproxy-0.4.11-r2[webkit]" [ebuild]) (dependency required by "net-libs/glib-networking-2.40.1-r1[libproxy]" [ebuild]) (dependency required by "net-libs/libsoup-2.46.0-r1" [ebuild]) (dependency required by "media-plugins/gst-plugins-soup-0.10.31-r1" [ebuild]) (dependency required by "media-libs/phonon-gstreamer-4.7.2[network]" [ebuild]) (dependency required by "media-libs/phonon-4.7.2[gstreamer]" [ebuild]) (dependency required by "kde-base/kdelibs-4.13.2-r1" [ebuild]) (dependency required by "kde-base/khelpcenter-4.13.2" [ebuild])
How can I cleanly upgrade the system to a newer kernel, latest versions of packages (incl. ZFS support), without breaking it? I already removed udev and upower, and emerged eudev and upower-pm-utils.
I am willing to remove all packages mentioned above, that seem to block the update process, and re-emerge them later. How wrong can it be to unmerge, for example, kdelibs?
-
Using the config.py listed in the previous post, and the content of /boot/kernels:
ls /boot/kernels/3.12.22-KS.01/ System.map config initrd vmlinuz
That's what bliss-boot returns:
--------------------------- Bliss Boot - v1.1.0 Jonathan Vasquez <jvasquez1011@gmail.com> Distributed under the MPL 2.0 --------------------------- Scanning /boot/kernels ... Scanning /etc/bliss-boot/config.py ... Please add your desired kernels and their options to the 'kernels' list in /etc/bliss-boot/config.py. These entries should match the kernels you have in /boot/kernels.
What is missing?
How to "adjust" euse".
in Desktop Help
Posted
Works for me!