Installed software: diaries
1. (12.06.13) System installation (Fedora 17)
General remarks
- The system was installed from a Fedora-17-x86__64_DVD disc. It was a graphical installation in the
Troubleshooting -> Install in a basic graphics mode
branch (normal branch lead to a hangup of the PC).
- Unfortunately, the screenshots were not found in the
~root/anaconda-screenshots
,
where they had to be.
This can be associated with a bug discussed in AFTER all the installation (including the "firstboot" stage) had finished, I found only one file
screenshot-0052.png
in the
/tmp/anaconda-screenshots
directory which I saved to
~root/anaconda-screenshots
- As a guide to the installation, I used my notes in the
-
Inst_SW_aux/F17_Installation_Guide_Notes.html
( local copy, requires the character encoding to be set to either Western Windows-1252 or Cyrillic Windows-1251 ) .
Step 1. Loading system from disc
Hard disk partitioning.
Two Windows partitions were available:
sda1
sda2 (~100Gb, "Win7") .
The following new partitions were created:
sda3 250Mb ext4 /boot
sda4 extended partition
sda5 176GB LVM physical volume
sda6 250Mb ext4 /boot2 a reserve for future Linux systems
sda7 250Mb ext4 /boot3 -"-
sda8 250Mb ext4 /boot4 -"-
The
/boot[2-4]
were specified
BEFORE the LVM PhysVolume and first
got natural names
sda[5-7]
, but specifying the LVM volume
moved them
to
sda[6-8]
for an unknown reason.
Here is a post-installation report on the the above partitions
by
fdisk:
[root@localhost Install]# fdisk -l /dev/sda
Disk /dev/sda: 750.2 GB, 750156374016 bytes
255 heads, 63 sectors/track, 91201 cylinders, total 1465149168
sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x85dc822c
Device Boot Start End Blocks Id System
/dev/sda1 * 2048 206847 102400 27 Hidden NTFS WinRE
/dev/sda2 206848 195518463 97655808 7 HPFS/NTFS/exFAT
/dev/sda3 195518464 196030463 256000 83 Linux
/dev/sda4 196030464 1465147391 634558464 5 Extended
/dev/sda5 196034560 556482559 180224000 8e Linux LVM
/dev/sda6 556484608 556996607 256000 83 Linux
/dev/sda7 556998656 557510655 256000 83 Linux
/dev/sda8 557512704 558024703 256000 83 Linux
LVM partitioning.
One LVM volume group (phys. extent size 4Mb) was created named
vg01
with the following LVM logical volumes:
lv_swap 4Gb
lv_root1 16Gb ext4 /
lv_opt1 20Gb ext4 /opt
lv_x01 50Gb ext4 /x01
occupying 90 Gb. The rest free ~86 Gb of the LVM phys. volume
are left for, say, 16 Gb (
lv_root2) + 20 Gb (
lv_opt2) of the
next Linux (Fedora 18 ?) plus 50 Gb of a general reserve.
Post-installation reports on partitions and devices
/etc/fstab
Here is the description of all the (standard + LVM) partitions
in the post-installation
/etc/fstab:
# /etc/fstab
# Created by anaconda on Wed Jun 12 06:53:53 2013
#
# Accessible filesystems, by reference, are maintained under
# '/dev/disk'
# See man pages fstab(5), findfs(8), mount(8) and/or blkid(8) for
# more info
#
/dev/mapper/vg01-lv_root1 / ext4 defaults 1 1
UUID=a984859d-099b-4fc7-a5bd-c0330eee186f /boot ext4 defaults 1 2
UUID=adefa6ae-ccda-4c58-8f26-214fef90236e /boot2 ext4 defaults 1 2
UUID=eea0a12e-a855-4327-9ac6-4d79d1659c3f /boot3 ext4 defaults 1 2
UUID=a1b1e9df-dbb0-4dad-9f55-cf2c359ebe60 /boot4 ext4 defaults 1 2
/dev/mapper/vg01-lv_opt1 /opt ext4 defaults 1 2
/dev/mapper/vg01-lv_x01 /x01 ext4 defaults 1 2
/dev/mapper/vg01-lv_swap swap swap defaults 0 0
df -h
Another table of (post-installation) device - mount point
correspondence is given by the
`df -h' command:
[fedotov@localhost]~% df -h
Filesystem Size Used Avail Use% Mounted on
devtmpfs 2.0G 0 2.0G 0% /dev
tmpfs 2.0G 92K 2.0G 1% /dev/shm
tmpfs 2.0G 2.7M 2.0G 1% /run
/dev/mapper/vg01-lv_root1 16G 9.5G 5.1G 66% /
tmpfs 2.0G 0 2.0G 0% /sys/fs/cgroup
tmpfs 2.0G 0 2.0G 0% /media
/dev/sda3 239M 88M 139M 39% /boot
/dev/mapper/vg01-lv_opt1 20G 242M 18G 2% /opt
/dev/sda6 239M 2.1M 224M 1% /boot2
/dev/sda8 239M 2.1M 224M 1% /boot4
/dev/sda7 239M 2.1M 224M 1% /boot3
/dev/mapper/vg01-lv_x01 48G 409M 46G 1% /x01
blkid
Another list of devices:
[root@...]# blkid
/dev/sda1: LABEL="System [Don't Touch]" UUID="F828C44D28C40C96" \
TYPE="ntfs"
/dev/sda2: LABEL="Win7" UUID="C2E6C562E6C55777" TYPE="ntfs"
/dev/sda3: UUID="a984859d-099b-4fc7-a5bd-c0330eee186f" TYPE="ext4"
/dev/sda5: UUID="0Yptz3-QJ1d-qhg0-HEqa-nD9p-jKse-RUy1Jc" \
TYPE="LVM2_member"
/dev/sda6: UUID="adefa6ae-ccda-4c58-8f26-214fef90236e" TYPE="ext4"
/dev/sda7: UUID="eea0a12e-a855-4327-9ac6-4d79d1659c3f" TYPE="ext4"
/dev/sda8: UUID="a1b1e9df-dbb0-4dad-9f55-cf2c359ebe60" TYPE="ext4"
/dev/mapper/vg01-lv_root1: UUID="2ae2f462-6fe5-4427-b8e6-faa7dcaf004d" \
TYPE="ext4"
/dev/mapper/vg01-lv_swap: UUID="be1bcd3a-b9d6-450f-8d99-98aa44092e9f" \
TYPE="swap"
/dev/mapper/vg01-lv_x01: UUID="b7e9002e-ef88-48eb-b0a7-88cdb9bdebc8" \
TYPE="ext4"
/dev/mapper/vg01-lv_opt1: UUID="c820e6bc-3c3a-4057-8b8c-bab76258970b" \
TYPE="ext4"
Network parameters
- We did not configure any Network parameters. A wired ethernet connection to an ONT modem (mgts GPON) was recognised and configured automatically under the name of p2p1 .
Boot loader
- The boot loader (GRUB2) info was recorded into the /boot partion.
Package selection.
- All packages (including those optional) in all categories were selected: 2766 packages in total. The list is available in
-
/opt/root/Install/rpm_history/130617.0208.rpm-qa.Initially_installed
Step 2. Adding a Fedora-17 entry to Windows 7 bootloader menu
Step 3. Firstboot
Defining /usr/local and /home
- In the beginning, we entered a shell console with Ctrl+Alt+F2 and made some renaming:
Creating account fedotov
- The account fedotov was created with
(user id, group id) = (501, 501)
as it was in the fedora 10 @ dom2 installation (purpose: to have a correct user/group for fedotov's files on the internal and external disks of dom2).
This violated the recommendations of F17 to have ID's > 999, and placed the account into a user/group "system" area, such that it is not seen by default in the lists of the system-config-users
utility (the "hide system entries" option in the settings should be deselected in order to make the account fedotov visible).
Post-installation file-system occupancies (Mon Jun 17 01:57:22 MSK 2013)
-
[root@localhost ~]# df -h
Filesystem Size Used Avail Use% Mounted on
rootfs 16G 9.7G 5.2G 66% /
devtmpfs 2.0G 0 2.0G 0% /dev
tmpfs 2.0G 96K 2.0G 1% /dev/shm
tmpfs 2.0G 29M 1.9G 2% /run
/dev/mapper/vg01-lv_root1 16G 9.7G 5.2G 66% /
tmpfs 2.0G 0 2.0G 0% /sys/fs/cgroup
tmpfs 2.0G 0 2.0G 0% /media
/dev/mapper/vg01-lv_opt1 20G 570M 18G 4% /opt
/dev/mapper/vg01-lv_x01 49G 1.2G 46G 3% /x01
/dev/sda7 248M 12M 224M 5% /boot3
/dev/sda6 248M 12M 224M 5% /boot2
/dev/sda8 248M 12M 224M 5% /boot4
/dev/sda3 248M 49M 187M 21% /boot
/dev/sda2 94G 59G 35G 63% /media/Win7
The most interesting lines are:
rootfs 16G 9.7G 5.2G 66% /
/dev/mapper/vg01-lv_root1 16G 9.7G 5.2G 66% /
/dev/sda3 248M 49M 187M 21% /boot
2. (17.06.13) /x01/legacy/ directory for dom, dom2, ... stuff
- The following subdirs in /x01/legacy/ created:
- → dom/RH9/ →
opt_root_Install -- a "cp -rp" of
dom2:/opt/root/dom_legacy/opt_root_Install
usr_local -- a "cp -rp" of
dom2:/opt/root/dom_legacy/usr_local
- → dom2/F10/ →
home/fedotov -- for selected stuff from dom2:/home/fedotov
opt ->
root/Install -- a "cp -rp" of
dom2:/opt/root/Install
usr_local -- a "cp -rp" of
dom2:/opt/usr_local
3. (17.06.13) ~fedotov/.emacs, ~root/.emacs
- copied from
~fedotov/.emacs
at dom2
(via /x01/legacy/dom2/F10/home/fedotov/.emacs
);
the original F17 version moved to .emacs.original_F17_version
:
- ~fedotov:
mv .emacs .emacs.original_F17_version
cp -rp /x01/legacy/dom2/F10/home/fedotov/.emacs .
[fedotov@localhost]~% ll .emacs
-rw-r--r--. 1 fedotov fedotov 3505 Aug 26 2010 .emacs
- ~root:
cp -rp /x01/legacy/dom2/F10/home/fedotov/.emacs .
chown root.root .emacs
[root@localhost ~]# ll .emacs
-rw-r--r--. 1 root root 3505 Aug 26 2010 .emacs
4. (18.06.13) Fisrt SYSTEM UPDATE; looking at Apper working
- The update is done via the Apper tool ("Get, Remove and Update SW", accessible via f -> Applications -> Administration -> SW Management) by
- selecting Updates;
- clicking "Check for new udpates" -> 1268 packages found;
- clicking "Apply" -> 150 additional packages to be installed;
- waiting about 70' for the procedure to finish;
- rebooting the computer (there was a clickable icon of a a message window ~"restart required" followed by a list of names of 15-30 packages, the window popping up when the mouse was brought over the icon).
- The total amount of installed packages is now 2904 (was 2766 initially).
The list is in
-
rpm_history/130619.0016.rpm-qa.After_first_update
5. (20.06.13) Trying to switch on the nouveau driver i.o. vesa in xorg.conf
- F17 comes with the nouveau driver for nvidia cards:
[fedotov@localhost]/etc/X11% rpm -qa |grep nouveau
xorg-x11-drv-nouveau-0.0.16-37.20120306gitf5d1cd2.fc17.x86_64
[fedotov@localhost]/etc/X11% lsmod |grep -i nouv
nouveau 941736 0
mxm_wmi 12865 1 nouveau
wmi 18697 2 mxm_wmi,nouveau
video 18991 1 nouveau
i2c_algo_bit 13257 1 nouveau
drm_kms_helper 46708 1 nouveau
ttm 79629 1 nouveau
drm 264481 3 ttm,drm_kms_helper,nouveau
i2c_core 38353 6 drm,i2c_i801,drm_kms_helper,\
i2c_algo_bit,nouveau,videodev
- In the article "Troubleshooting Nouveau"
(http://nouveau.freedesktop.org/wiki/TroubleShooting/
)
one finds the Basic Question 4:
- "Is X using the right driver (DDX)?"
- "Do not use `nv` driver in X. Use `nouveau` instead.
If you do not explicitly tell X to use `nouveau` in `xorg.conf`
,
it will usually default to `nv`, or maybe even `nvidia`
(the proprietary driver). Any X driver (e.g. `vesa`) other
than `nouveau` or `fbdev` will cause problems with KMS."
- As our our
xorg.conf
does not contain the `nouveau`:
[fedotov@localhost]/etc/X11% cat xorg.conf
Section "Device"
Identifier "Videocard0"
Driver "vesa"
EndSection
we tried to change vesa -> nouveau and reboot.
Fortunately, the boot screen worked OK, but X's did not load ...
We could open a root console via Ctrl+Alt+F2 and change xorg.conf
back to the original status.
6. (20.06.13) xx, xxx scripts
7. (20.06.13) Trying to switch on the nouveau driver io vesa with xorg.conf generated by Xorg
- In section 5, we failed by changing vesa -> nouveau in the original
/etc/X11/xorg.conf
which is a very short one.
- Now we try to generate an
xorg.config
with Xorg .
Step 1 : Xorg :1 -configure
- This indeed created the
xorg.conf.new
(see here) which we placed into
Inst_SW_aux/130620.Xorg-configure
directory.
Step 2. Editing the generated conf file
Despite the message
"Configuration failed", we try to continue.
The
xorg.conf.new
containes:
3 Monitor sections
3 Device sections corresponding to driver/Id combinations
nouveau / Card0 ,
fbdev / Card1 ,
vesa / Card2 .
respectively,
3 Screen sections
We commented out the
second and
third Sections for
Monitor,Device and
Screen (also left only
one Screen in the Section
"ServerLayout")
and put the result into
xorg.conf.new.edited
(see
here)
Step 3: Try xorg.conf.new.edited
- There are also files
boot.log
(see here)
session_last_screen
(see here) -- a correctly formatted end of boot.log
Step 4: Restore the original /etc/X11/xorg.conf
and reboot
- Note that we keep a copy of the original file in
/etc/X11/xorg.conf
7B. (21.06.13) Boot without /etc/X11/xorg.conf
- So we tried this mode. X started OK, but in the same 1024x768 mode as earlier. The Xorg log file is in
Inst_SW_aux/130621.Xorg.0.log.booot_without_xorg_conf
(see here)
- The original
xorg.conf
was restored after the exercise.
8. (22.06.13) Try nouveau driver with /dev/dri/card0 created manually
In sections 5 and 7, the attemts to start the
nouveau video driver
failed formally due to an absence of the
/dev/dri/card0, .../card15
files. What if we create the
/dev/dri/card0
manually?
how to add /dev/dri/card0 ?
- The man page above contains the following text (absent in the F17 "man drm"):
"X(7) will attempt to create the device node automatically.
To create the device node manually:
mkdir -p /dev/dri
mknod /dev/dri/card0 c 180 0
chgrp wheel /dev/dri/card0
chmod 0660 /dev/dri/card0
To enable DRI in the X configuration add the following to either
xorg.conf for Xorg(1) or XF86Config for XFree86(1) :
Section "Module"
...
Load "dri"
Load "dri2"
Load "glx"
EndSection
...
Section "DRI"
Group "wheel"
Mode 0660
EndSection
"
- Why not to try this for the F17 nouveau driver?
- One question is about the group `wheel' which is present in F17 but w/o any Group Member :
- leave the file in the `root' group but set the mode 0666 i.o. 0660 permitting the rw access to everyone?
- Another question is
- what should be the the last two parameters of the `mknod' command in our case?
- They are MAJOR and MINOR device numbers.
- google search `how to find major and minor for mknod fedora 17'
- -> http://timocharis.com/help/udev.html
"So where did I get the major and minor numbers? You'll find them
in your kernel source directory under Documentation/devices.txt
-- you have a recent kernel source handy, right? If not, you can
score at http://www.kernel.org/pub/linux/docs/device-list/
A caution: the device itself will be the first number listed.
So if you search the device list for /dev/sd and find sdb, you'll
note its major number is block 8 (that's fine) and the minor
numbers are 16-31 (because sdc starts with 32). However, the
device itself (/dev/sdb) has the minor number 16; the first
partition (dev/sdb1) gets the minor number 17. The second
partition (dev/sdb2) gets 18, etc. An off-by-one error is
entirely too possible here.
"
- -> the url http://www.kernel.org/pub/linux/docs/device-list/
does not exist!
- -> google search: `linux device-list'
- -> http://en.wikipedia.org/wiki/Linux_Assigned_Names_and_Numbers_Authority
- -> The Linux Assigned Names And Numbers Authority
http://lanana.org/
- -> The Linux Device List
- -> Current Linux 2.6+ Device List 06 April 2009
http://lanana.org/docs/device-list/devices-2.6+.txt
- -> find /dev/dri ->
"226 char Direct Rendering Infrastructure (DRI)
0 = /dev/dri/card0 First graphics card
1 = /dev/dri/card1 Second graphics card
...
"
- -> CAN THE NUMBERS 226,0 BE ASSUMED TO BE THE SEARCHED MAJOR,MINOR PAIR?
- -> searching for
/dev/sd
in the list, gives
"8 block SCSI disk devices (0-15)
0 = /dev/sda First SCSI disk whole disk
16 = /dev/sdb Second SCSI disk whole disk
32 = /dev/sdc Third SCSI disk whole disk
...
240 = /dev/sdp Sixteenth SCSI disk whole disk
"
which nicely agrees with the second paragraph of the quotation from http://timocharis.com/help/udev.html
above.
So the assumption may well be valid!
- -> google search `"mknod /dev/dri/card0 c 226 0"' gives many hits
- -> e.g.
http://forums.freebsd.org/archive/index.php/t-27428.html
in the following context (quite reminding our case):
"173. drmOpenDevice: node name is /dev/dri/card0
174. Failed to change owner or group for file
/dev/dri! 2: No such file or directory
175. Failed to change owner or group for file
/dev/dri/card0! 2: No such file or directory
176. drmOpenDevice: open result is -1, (No such file
or directory)
. . .
The device node for card0 isn't created.
Try starting X after creating it manual.
Please search for it under /dev, if it's there,
link to it.
If it isn't there, create a device node. mknod
mknod /dev/dri/card0 c 226 0
This is a temporal workaround. The node will be
cleaned out after a reboot, but let's see
if X can be started.
"
The warning "THE NODE WILL BE CLEANED OUT AFTER A REBOOT" should be noted !?
Trial 1
-
mkdir -p /dev/dri
mknod /dev/dri/card0 c 226 0
with the vesa driver
- reboot -> started as usual;
/dev/dri/
directory disappeared!
- May be the ownership by the group wheel would save the file?
Trial 2
Trial 3
- No hope to preserve the
/dev/dri/card0
card, but, nevertheless, try as in Trial 2, but with vesa->nouveau replacement in /etc/X11/xorg.conf
9. (25.06.13) System Update
- Update done with Apper and Fedora repos.
- 34 rpm's have been updated.
Details in:
rpm_history/130625.0059.rpm-qa.After_sys_update
(see here)
- The Apper got broken in the end.
- First it said something like "module setting changed, want to update?"
- I answered "Yes". Then it started to infinitely show "Waiting for other task...". I cancelled this finally, then exited answering "No" (or was it "Discard"?) on the same question on changed settings...
- When started next time, it can not finish a Check for new updates;
also can not show the list of Installed sw...
- After a reboot the Apper is OK.
- The reason for the malfunctioning MAY have been an update of the
selinux-policy
package (3.10.0-169 -> 3.10.0-170) ?
Tips