Saturday, January 7. 2012
As an optimist I decided to upgrade my F15 installation to F16 installation. As a paranoid person I made a backup. I started at noon and I am still doing the upgrade that now has progressed to be a text based reinstall that has wiped my old system completely. Here is how.
First I have two displays, both are wide screen displays but have different resolutions. And the "bigger" display is connected with HDMI and the smaller with DVI. This means that when the graphical upgrade boots it results a screen where the buttons are somewhere "under" the both displays. As an optimist I decided to continue anyway and was able to guess how to "tab" between buttons to get it going. SELinux failed or didn't, can't tell as the screen was too small and the progressbar stopped at 100%. A reboot and another try with "text" upgrade produced grub shell.
Affer some goofing with rescue I managed to configure grub2 to try to boot the system but alas the installed kernel immediately fails. Booting with old F15 kernel brings the system up. Playing around with yum update shows that the yum thinks that the system is F15 not F16 althoug the relese-file claims otherwise. With --releasever 16 I get a kernel that installs but the dependencies are completelely borked. Despite --fix-broken and other options the battle is finally lost. Ofcourse I believe this after a good fight and some googling.
Then I decide to do a complete reisntall having a small recollection that at least the graphical partition tool lets me to modify the scheme - that is the graphical tool not the curses based tool. Whoops I no longer have any installation left. The right way would probably have been either disconnecting the other display or finding out if there is geometry option for the graphical installer. Then I might have been able to save the home, backup and opt volumes. Luckily I had backed almost everything to NAS as well.
The result with text-based installer is quite bare and you need to know how to start network manually to continue. One way to do it is to "up" the network device. In my case the eth0.
ifup eth0
Then I did the first "yum update" and while it was running fixed the volumes. I resized the home and did a new opt volume.
lvm
lvresize -L 100G VolGroup/lv_home
lvcreate -L 50G -n opt VolGroup
quit
These should have been done after unmounting /home. I made fstab-entry for opt and filesystems for both:
mkfs.ext4 /dev/mapper/VolGroup-lv_opt
yum grouplist > groups.txt
vi groups .txt
while read i;do yum install -y "$i";done < groups.txt
cd /etc/systemd/system
rm default.target
ln -s /lib/systemd/system/graphical.target default.target
shutdown -r now
To get the google-chrome running:
sudo chcon -t textrel_shlib_t /opt/google/chrome/chrome-sandbox
Both gconf-editor and gnome-tweaktool should be installed. The latter is used to give the desktop to Nautilus and the former to add virtual desktops to secondary displays as well.
Wednesday, July 20. 2011
The two latests updates on Gnome that I have suffered have greatly reminded me of the famous Linus quote
"This "users are idiots, and are confused by functionality" mentality of Gnome is a disease. If you think your users are idiots, only idiots will use it. I don't use Gnome, because in striving to be simple, it has long since reached the point where it simply doesn't do what I need it to do."
And I also remember him switching to Gnome when KDE 4.0 hit the fan. I was more than tempted to switch to Windows 7 after my first install of F 15 and having tried KDE with F 14. I was so disgusted with both "UI"s that I switched to MAC at home but couldn't at work as it is not supported on the hardware that is blessed by those who know. better. I griped with F 14 Gnome and finally grew into it. At home I found that Mac OS X gui was a bit better on some tasks and sucked more in some other tasks.
After initiial installation nothing worked at work with F 15 as I expected but the lure of first working ALT+TAB switcher finally won me over. Of course this involved a lot google-fu to fix things that really annoyed me:
- Desktop should have Icons. How to do this is blogged over and over again. One quite short version of the instructions can be found here.
- If you have two monitors you probably want workspaces on the other screen as well. Again gconf-editor to the rescue. Instructions can be found here
- Custom launchers. It seems that you can use Alacarte to create the launchers. Unfortunately Alacarte creates the desktop files without Categories and the launchers end up in the "Muut" -section (Others). It seems that you need to conjure the gedit up and add the categories by yourself. Files are located under your home directory in ".local/share/applications" (Note the dot at the beginning of the path). Just add Categoris=Game if your custom app is a game.
- Weather. There was no weather on my taskbar! Now there is with gmone-extension.
Not gnome related annoyances:
- Youtube (in some sense == flash). With my laptop the only way to get it to work was to install gnash-plugin and on my desktop the only way to it to work was to install latest beta from Adobe.
- The lack of the usual stuff after install. This time I referred several sources Fedora setup guides.
- Gnucash. It is shocking to find out that you owe 3k euros instead of having 2k euros on your savings account. And it seems that the error is gradual. It is time to say goodbye forever for this piece of shit as soon as I have time to find better alternative. Meanwhile I will continue to use it on Mac and will not update it. It is truly sad that the history will be gone forever as soon as I bury the mac. I guess it will not be Lion for me for a quite some time.
-
Thursday, October 28. 2010
The system comes with prepartioned disk - four primary partitions, where windows 7 has been crammed to one small 200G partition.
The first partition is 300MB and contains something in NTFS, second partition contains the forementioned windows 7, third is 15G recovery partition and the fourth is 2G FAT32 tools partition.
The new scheme will have shrinked winodws 7 partition and no 2G FAT32 partition. Shrinking is done through windows and so is the obliteration oft he fat-partition. I just simply copied the contents of the tools partition to the windows 7 partition before deleting it.
Windows 7 is a huge beast, after first shrinking it still required a partition of 113,61G! There a files that the "shrinker" can not move because they are used by the operating system. There is still 74G free space on the disk. Well, I will tackle this later.
I've downloaded the gPXE iso-image, burned it to a cd and will use it as the installation medium. The installer crashes on the graphical mode, so the key here is to use "text"-option or "vnc" to get it going.
First installation completed while I was having lunch. Quick test with "init 5" confirmed that display works and installation of 64bit flash-plugin confirmed that sounds and network work. Did a resize for the for the ntfs filesystem and forgot to resize the partition. Boot to the windows 7 confirmed that it still worked. Second boot to linux and deleting + recreating the partition with smaller size caused a bluescreen. The reason for this was probably that I used parted to nuke the partition and recreate it which nuked something else than partition table or changed the starting location of the new partition. Mental note, next time use fdisk instead! Repair with installation cd managed to fix the boot problem.
The second install of Fedora 14 beta is necessary as the free space I got from the second shrinking is between the extended partition and the shrinked Windows 7 partition. The limit for primary partitions is 4 and this means that the extended partition should be resized. I don't know any tool to do this so I just start from the beginning and nuke the old extended partition on the reinstall of Fedora 14. Mental note: next time I will use live cd to resize the windows-filesystem and partition.
The second install went as smoothly as the first one. There was some network traffic so I managed to get bored while waiting the packets to load. Some after installation steps:
- Create a user with -G wheel for myself
- Edit /etc/inittab - replace default boot level from 3 to 5
- visudo - give wheel group super user
- Install 64bit Flashplugin by surfing to youtube and following links. Did the personal install - as there will be no other users on this laptop than me - which means uncompressing the plugin to .mozilla/plugins folder under home folder.
- Install 32bit wine as not trusting it to work with spotify. Spotify works with 32bit wine.
- Freennx-client is still useless.
- avahi-daemon is not able to resolve anything; sudo service avahi-daemon stop; sudo chkconfig avahi-daemon off
- sshmenu-applet requires logoff/logon.
- komodo won't install without setenforce 0 and
might have serious trouble with .gvfs-directories or the problem is within gvfs itself. Remains to be tested. Komodo edit-6.0 is not a good idea with .gvfs-directories it clears the contents of the files while saving. Lost two files with it, luckily subversion came to rescue and only the edits where lost. Same goes with Eric and Eclipse, the only editor so far to handle .gvs-files is gedit.
- Remmina is great!
- Did not manage to use two external displays with docking station - at least so that one display is in DVI and one is in VGA connector. The Dell display on the VGA connector did not get recognized and got resolution it claimed it could not handle. Probably this is not supported and displays should be connect with some other connector instead. Remains to be seen. Also two different displays is probably not a good idea. Now I have the laptop display and one external display on DVI connector.
- Bluetooth works - I guess - I was not able to browse my Galaxy S phone. Or send files to it but I was able to receive a file from it.
And android notifier ping did not get through. It almos works with ping and some battery events got through. So, it works as badly as it did with previous version of Android notifier.
- Fingerprint reader not supported by the driver. Perhaps next year - at least on some Ubuntu
Monday, May 3. 2010
There are other installation guides: CentOS 5.2/64bit, CentOS5.4 and one from Magento forum. This is another take on the matter.
- Install 32bit server-gui CentOS 5.4, with SELinux on permissive
yum update
yum install mysql-server, and remember set the password as instructed! Also starting the server and configuring it to start on boot might be a good idea.
- Install ZendServer
- Create a file
/etc/yum.repos.d/zend.repo with*
yum clean all
yum install zend-server-ce-php-5.3
/usr/local/zend/gui_passwd.sh
/usr/local/zend/bin/zendctl.sh restart
- create mysql-database for magento:
mysqladmin -u root -p create magento
- connect the databas
e mysql -u root -p magento
- create user
: create user 'magento'@'localhost' identified by 'yourpasswordhere';
- grant privileges:
grant all privileges on magento.* to 'magento'@'localhost' with grant option;
- Fetch the magento package
- untar it to the /var/www/html
- Set permissions: setfacl -R -m user:apache:rwx media var app/etc/
- Browse to your server/magento to start the installation. The only catch now is the socket, it is /var/lib/mysql/mysql.sock
*
[Zend]
name=Zend Server
baseurl=http://repos.zend.com/zend-server/rpm/$basearch
enabled=1
gpgcheck=0
[Zend_noarch]
name=Zend Server - noarch
baseurl=http://repos.zend.com/zend-server/rpm/noarch
enabled=1
gpgcheck=0
Sunday, April 25. 2010
It seems that it has become impossible to run web-services on Debian 64bit platform. The segfaults are now too frequent and repeatable on the same web page. There is something completely broken on Lenny. I even tried using Zend Server instead of regular php and the results are still:
[Sun Apr 25 11:57:37 2010] [notice] child pid 5240 exit signal Segmentation fault (11)
[Sun Apr 25 11:57:38 2010] [notice] child pid 5241 exit signal Segmentation fault (11)
[Sun Apr 25 11:57:38 2010] [notice] child pid 5242 exit signal Segmentation fault (11)
[Sun Apr 25 11:57:39 2010] [notice] child pid 5214 exit signal Segmentation fault (11)
[Sun Apr 25 11:57:40 2010] [notice] child pid 5218 exit signal Segmentation fault (11)
[Sun Apr 25 11:57:40 2010] [notice] child pid 5235 exit signal Segmentation fault (11)
[Sun Apr 25 11:57:41 2010] [notice] child pid 5216 exit signal Segmentation fault (11)
I guess I've also found this happy camper:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=537794 and it is the end of the road for me.
Also the second "happy happy joy joy"- moment was that with newer Zend-Framework. The old code stops working completely with "Fatal error: Unsupported operand types in /usr/local/zend/share/ZendFramework/library/Zend/Translate/Adapter.php on line 212". No trace but I guess it was caused by the very first line in the code. Downgrading and manually removing all cache-files helped with this but the segfault remains.
The segfault also happens on CentOS as well and it is not actually related to fore mentioned Debian bug at all, even though I have that bug as well. Somehow I had managed not to commit a change I had made and a regression bug popped up. This segfault was caused by by a method calling itself in an unending recursion. When the stack runs out, it segfaults the php which I'm actually very happy about.
Saturday, January 2. 2010
The only one thing I keep the original Asus EEEPC windows installation hanging around is the PC Suite. The only thing I do with it is to back the phone up. This means that when I do boot to the Windows the vicious cycle of rebooting starts. What makes this experience even more gratifying (for the masochists) is the Norton Security and the stability and correctness of the PC Suite, the phone and the updater software.
I started yesterday and I am still trying to succeed.
- I have booted the laptop now seven times. And the windows is still updating.
- The PC Suite has updated itself once and required a boot
- Norton has expired and it has been uninstalled, that required a boot
- The Nokia Software updater has updated itself and required a boot
- After three boots the computer has lost the connection with the phone and wanted to run the connection wizard and reinstall drivers that are already installed
- Seven tries for the software updater have had following results
- 3 times no connection to the internet
- 1 time identified the phone and found update to install but the whole thing crashed after that
- 2 times "Service not available"
- 1 time direct crash and windows wanting to send bug report to Microsoft
I have been thinking of buying the N900 phone to replace my current phone but if it requires use of PC Suite for anything my next phone will not be a Nokia.
While writing this the eight reboot has finished and the windows is still updating. And it has now finished and requires the ninth reboot!
Monday, December 7. 2009
Dec 7 19:12:38 vcust362 kernel: [465884.894733] php[22592]: segfault at 7fca2b160ac0 ip 7fca2b160ac0 sp 7fff2b6c4b68 error 14 in libtasn1.so.3.0.15[7fca3009d000+f000]
Yksinkertainen tosiasia taitaa olla, että 64bittisien Lennyn php ei toimi. Enkä tiedä korjaisiko asian se, että kääntäisin itse uuden version. Tämä on kertakaikkisesti liian työläs siirtyminen toimivasta CentOS ympäristöstä. Olen nyt hukannut 8 tuntia siihen, että olen yrittänyt saada edes toimivan ajoympäristön php+java kilkkeille. Mikään ei mene tappelutta, kaikki kaatuu ja segfaulttaa. Kaikki oletukset on väännetty ja käännetty varoiksi vielä niin, että suurin osa yleisistä web-ohjeista on väärässä, kun se debianin tapa vaan tehdä se juttu on niin oikein ja kaikki muut ovat väärässä.
Sunday, December 6. 2009
Debian always fails when you need it most:
open("/usr/share/locale/en_US.UTF-8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory)
open("/usr/share/locale/en_US.utf8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory)
open("/usr/share/locale/en_US/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory)
open("/usr/share/locale/en.UTF-8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory)
open("/usr/share/locale/en.utf8/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory)
open("/usr/share/locale/en/LC_MESSAGES/libc.mo", O_RDONLY) = -1 ENOENT (No such file or directory)
write(2, "make: "..., 6make: ) = 6
write(2, "*** [yritykset.imp] Segmentation "..., 38*** [yritykset.imp] Segmentation fault) = 38
write(2, "\n"..., 1
It is nice that installed new operating system, installed by the professionals of the virtual machine provider is broken from the very first meters. I haven't yet met Debian installation that works as specified and Debian is still the only linux distribution I am really afraid to upgrade. Especially if the upgrade contains core pieces and require a boot. Debian has actually wasted more of my time than windows ever has! To be honest, the reason for that is that I am more capable of tracing errors and fixing things with Debian than with windows. I am not so sure anymore that this is a good thing. I am usually smart enough to give up with windows in time but Debian being kind of linux is just begging it. And of course I can always bitch and moan afterwards...
Monday, November 30. 2009
- Download and untar as usual
- Modify installer.sh and comment out architecture-checks*
- sudo yum install gtk2-devel-2.18.3-21.fc12.i686
- sudo yum install compat-libstdc++-33-3.2.3-68.i686
- sudo yum install compat-expat1-1.95.8-6.i686
- run installer.sh
In case installer complains something that is still missing apply pattern:
- sudo yum provides '*libexpat.so.0*', where replace libexpat.so.0 between asterisk with missing thingie
- sudo yum install compat-expat1-1.95.8-6.i686, where replace compat-expat1-1.95.8-6.i686 with result of previous step. Note that there will be several hits. Pick the most promising one with ending .i686, that is the 32bit version...
- run installer.sh
Resulting installation may or may not work, I give no guarantees.
* Section commented out:
# Show error if we are not running on x86
#case `uname -m` in
# i?86) ;;
# *) echo "This installer only has x86 binaries. Sorry.";
# exit 1;;
#esac
Tuesday, November 24. 2009
Easy. Allow users to install packages without root-password. That means that you can install trojans with root-privileges. Meanwhile add very well tested (NOT!) selinux-rules that make life terribly hard. And update the management commands so that they are not backward compatible with previous syntax.
For example let's say that you try to install Komodo-edit to do some php-work? It will not work. Let's say that you have donwloaded ZendFramework and installed it to your own-directory. It will not work and even after the instructions given by the selinux to fix the problem you will finally run into a situation where selinux is preventing you to use the fix command given by selinux.
What is the result? You have a system that will have selinux disabled and root access for package installation without password. FAILURE!
Thursday, October 22. 2009
I have CentOS 5.3 server with additional repositories, see extended entry for versions of ffmpeg and mplayer. I have written two scripts for playbacks. First one is:
#!/bin/bash
song=$1
bitrate=$2
#echo $song >> /tmp/songs.txt
/usr//bin/ffmpeg -i "$song" -f mp3 -ab $bitrate - 2>/dev/null
which I call swiss.sh. This is from Activeobjects-forum.
And the second one called wma.ssh is:
#!/bin/bash
song=$1
mplayer -vo null -msglevel all=-1 -quiet -srate 44100 -channels 2 -vc dummy -ao pcm:nowaveheader:fast:file=/dev/stdout "$song" 2>/dev/null
Sorry, I've lost the original source. First one is used with just one step: "swiss.sh %s %b" and the second one requires two steps. First step: "wma.sh %s" and second "lame -r -b %b - -", the "-r" is essential!
Continue reading "Subsonic wma flac ogg playback"
Sunday, September 6. 2009
- Install iReport 3.6.0
- Do a report with it and let it compile the report and it's subreports for you.
- Fetch jasperreports-project 3.6.0 and uncompress it
- Run ant jar at the root of uncompressed directory
- Copy every jar-file from both dist- and lib-subdirectories to /var/lib/tomcat5/shared/lib
- Make sure that tomcat can read copied files
- Copy your database jdbc class to same directory
- Restart tomcat
- Do a directory that tomcat can write and read from
- Copy your compiled jasper-reports (.jasper)-fiels to that directory
Here is an example that can produce pdf-report with subreports and parameters from Postgresql-database.
<?php
require_once('java/Java.inc');
// Get the database connection
$drv = new Java('org.postgresql.Driver');
$objDbm = java("java.sql.DriverManager");
$objDbConnect = $objDbm->getConnection("jdbc:postgresql://host/databasename", "user", "passwords");
// My report has one subreport and requires five parameters
// they are all STRINGS! I tried dates and timestamps but something chokes
$company='1105366-9';
$toimpiste='kaikki';
$startDay="2009-08-01"
$endDay="2009-09-01"
//
// This is the directory tomcat needs to be able to write and read from
// it contains the compiled reports (.jasper-files)
//
$reportPath="/web/Pikkusisko2/raporttimallit/";
// Parameters are passed in java hashmap, one way to generate one is
// like this
$parameters=new Java("java.util.HashMap");
$parameters->put('pYritys',$company);
$parameters->put('pToimipiste',$toimipiste);
$parameters->put('SUBREPORT_DIR',$reportPath);
$parameters->put('pAlkupaiva',$startday);
$parameters->put('pLoppupaiva',$endday);
// Generate report pdf
$jasperRunManager = new Java("net.sf.jasperreports.engine.JasperRunManager");
$rep=$jasperRunManager->runReportToPdfFile("$reportPath/projektit2.jasper", $parameters, $objDbConnect);
?>
Sunday, May 31. 2009
Ubuntuporukoilla on oikein hyvät ohjeet. KPPP:tä en löytänyt ja NetworkManglerilla/graafisella kilkkeellä ei tullut suoraa voittoa, joten loin ohjeen kolme tiedostoa: elisa, elisa-connect-chat ja elisa-disconnect-chat tiedostot. Otin suoraan Elisan mallitiedostot ja vaihdoin bluetoothista kaapeliin. Fedorassa pon- ja poff-skriptat ovat saatavilla, mutta eivät ole polussa joten tein pari apuskriptaa:
#!/bin/bash
sudo service NetworkManager stop
base=/usr/share/doc/ppp-2.4.4/scripts
sudo cp /etc/resolv.conf.elisa /etc/resolv.conf
sudo $base/pon elisa
ja
#!/bin/bash
base=/usr/share/doc/ppp-2.4.4/scripts
sudo $base/poff
sudo service NetworkManager start
resolv.conf.elisaan kävin lätkimässä skriptan testiajon palauttamat nimipalvelimet:
nameserver 193.229.0.4
nameserver 193.229.0.42
Tuntuu toimivan.
Saturday, May 9. 2009
There is a thing called java-bridge that allows you, if I understood it correctly, to talk to either java runtime or mono runtime from php. And it also provides methods of doing that the otherway around. And if I understood it correctly there are two ways the php application can access the jvm, either by socket or by using application server. The access method is configured in /etc/php.d/java.ini.
Of course to get to configure anything you have to have java-bridge installed. In my case it meant downloading the latest source, yum installing tomcat and rpmbuilding the java-bridge package from the tarball with sudo rpmbuild -tb php-java-bridge_5.4.4.2.tar.gz. Build might require some additional packages to be installed, in my case everything was already in place. Build produced four rpm-packages:
php-java-bridge-5.4.4.2-1.i386.rpm
php-java-bridge-debuginfo-5.4.4.2-1.i386.rpm
php-java-bridge-devel-5.4.4.2-1.i386.rpm
php-java-bridge-mono-5.4.4.2-1.i386.rpm
I have so far installed php-java-bridge and php-java-bridge-mono. I have also run the java-bridge examples from tomcat and found out...
Update: 02-September-2009, this entry was interrupted by something probably related to leaking dishwasher which stopped this work for whole two months.
Wednesday, March 25. 2009
I was about to play a quick game of Ticket to Ride but alas after last evening's update the applet did not work. What was even more evil was to find out that nothing related to java functioned reliable.
After stracing I found that java hung on futex() and on read(). That was very warming and cozy but got me nowhere expect to Red Hat's bugzilla. And there the salvation was on comment 7. Now I just kill the applet before gaming and wait for next update which fixes imsettings applet.
After couple of updates this issue has gone away...
|