I don't understand. Eighteen months ago, I jumped into the PCLOS world and I'm very happy with what I've achieved so far. It hasn't all been smooth, as I wasted my first couple of weeks (on-off work time) learning about the rule of only a single repo
in /etc/apt/sources.list. However, I love PCLOS because of the forum and the distro is relatively easy to remaster. Since March of 2011, I've produced three generations of PCLOS live CDs (based on LXDE Mini), but the most recent generation (using the 2012.06 LDXE Mini) has provided a hellish trip. Part of my discomfort was due to the coincident loss of high-speed Internet access, as I began working on the new live CD. While I have a fully-operational, customized live CD based the most recent ISO
, I'm stuck with a couple of serious head-scratchers.
Since my last live CD remaster (from the beginning of 2012) went without a single hitch, I had no reason to suspect that the 2012.06 ISO would present any challenges. All of my remastered live CDs simply add a handful of packages I need to make PCLOS a better tool disc for PC repair, a portable media player, and an Internet access platform. Consequently, all of my changes can be characterized as cosmetic and require ZERO
manual changes to any of the PCLOS startup scripts or directory structures.
In spite of this, my first remaster attempts failed consistently due to the dreaded "failure to mount the squashfs file system error" at boot time. Eventually, I turned to the PCLOS forum and found my own first post, which was ultimately about the aforementioned misconfiguration of Synaptic (ie. too many repos selected in sources.list), which produced the same boot time error. With this remaster, I was certain I chose only a single repo, an explicit choice I made to try to compensate for the loss of high-speed Internet. I finally discovered that one of the pending updates was apt-sources-list
, which upon installation, changed the repo in mid-stream. Subsequent updates and/or new package installations (from a repo different from the one I started the remaster
) would somehow "corrupt" PCLOS and fail to boot as a live system. It took a lot of
hours for me unwind all of that. Not even sure if using two different repos (on completely different occasions) is problematic. I just eliminated yet another potential variable, to be on the safe side.
Let's create yet another VM on a PC with high-speed Internet access to test another hypothesis: to ensure a single repo was used from beginning-to-end. Upgrade and install new packages via Synaptic. Somehow, my Synaptic markings file omitted Java. Downloaded the related packages, made a live PCLOS ISO, and it failed as before. Performed a complete removal via Synaptic, remade the live PCLOS ISO, and was rewarded with a working PCLOS live system ! Can't explain it as I successfully ran the Java test suite on their (Oracle's) website before removing it from the live CD mastering VM. I wasn't confident of this conclusion when I started this message, so I reformatted the VM (again !) to reinstall/reconfigure my PCLOS. This time, I switched to using apt-get
to retrieve, cache, and install the .RPMs (won't be going back to Synaptic.) Scripting apt-get is a simpler, more reliable technique to ensure the same results across PCLOS installations:
apt-get -d -y install alsamixergui brasero chromium-browser draklive-install firefox flash-player-plugin grafburn gparted gpicview gsmartcontrol htop httrack java-1.6.0-sun joe libdvdcss2 partclone partimage vlc vlc-plugin-flac xarchiver
Now I have a complete set of RPMs from the heanet.ie repo, upgrades and optional RPMs, from a single
repo. The 1st ISO from this build failed ! Removing Java (via apt-get) yielded a (more-or-less) working live DVD. But there's more:
rm -vf pclos-lxde-mini-3.0b.iso pclos-lxde-mini-3.0b.iso.md5
mylivecd --verbose --splash=no --timeout=30 --usbhome --xz --ufs=unionfs pclos-lxde-mini-3.0b.iso ; md5sum pclos-lxde-mini-3.0b.iso >pclos-lxde-mini-3.0b.iso.md5
The last faulty ISO was under 250MB
! I removed Java, immediately remade the ISO (no other intervening actions besides running BleachBit), and the resultant ISO swelled to slightly beyond the capacity of a CD-R (705MB !) Go figure.
Here's the final PITA. I burned and tried this disc on a Toshiba Satellite M55-S331 laptop. Booted but never made it to the desktop, due to a silent X server death. The graphics chip is an Intel Mobile 915GM/GMS/910GML. (No, I didn't bother booting console mode and manually launching startx for more info.) During the months I evaluated PCLOS before jumping on the bandwagon, I'm sure I successfully tested it at least one laptop with this chip. Oh yeah, I still have my LXDE-Mini 2.1 disc and it works perfectly on the laptop. That's by far the biggest disappointment: failure on a platform that should
work, as I hoped to see improvement with respect to hardware detection and driver support, due to a newer kernel and other improvements. Can't always git what ya want....
Volume in drive E is DATA_2
Volume Serial Number is 3C88-BBEA
Directory of E:\Temp
07/07/2011 07:42 AM 694,171,648 pclos-lxde-mini-1.1.iso
06/24/2012 07:25 PM 613,959,680 pclos-lxde-mini-2.0.iso
09/22/2012 07:49 PM 604,239,872 pclos-lxde-mini-2.1.iso
09/25/2012 03:46 PM 608,067,584 pclos-lxde-mini-3.0-base.iso
09/27/2012 04:21 AM 693,233,664 pclos-lxde-mini-3.0-base1.iso
09/27/2012 11:58 AM 731,443,200 pclos-lxde-mini-3.0-base2.iso
09/29/2012 12:39 AM 737,890,304 pclos-lxde-mini-3.0-base3.iso
09/29/2012 01:04 PM 743,045,120 pclos-lxde-mini-3.0-base4.iso
10/09/2012 07:40 PM 739,676,160 pclos-lxde-mini-3.0b.iso
9 File(s) 6,165,727,232 bytes
0 Dir(s) 11,803,774,976 bytes free
Above are all of my working remaster projects. The 3.0-base ISOs are incremental improvements for testing purposes, to build confidence while I was trying to figure out what was wrong. I was hoping I'd be replacing 2.1 with 3.0, but it's time to move on to non-PCLOS projects, so that'll have to wait until at least the end of this month or next month. FWIW, this what I'll be carrying in my pocket:
To wrap this rather lengthy report, I can make a more-or-less working live CD based on a 3.2.18 kernel which bizarrely will not even boot
with Java (effectively an application !) installed. Is there any means of debugging this further ? Finally, what would possibly cause a fairly popular device with a decent history of support to fail ? Granted, the Satellite M55 isn't a new laptop (at least seven years old), but it really should work. By the way, my 3.0b disc wouldn't boot the X server via the FBDEV menu pick, which has almost never failed me in the past. The only occasions where FB mode has failed is with the very
latest laptops, hence my desire to have a live disc with a v3 kernel.