in my current situation, I have very few people to talk about linux woes, or even computer problems, and I rarely have such to get off my chest.
But, my wife practically killed my pc by cleaning our toaster "without getting it wet", and I had to figure out loads about various under the hood aspects of Linux / Mageia / KDE that I had never wanted to know about, that I just want to put this someplace.
Please feel free to stop reading this now, if my story doesn't interest you. I don't care for any comments in that direction.
Okay, so here it goes:
my hardware (server): Ryzen 7 Pro 5750GE APU, 32GB ECC-DDR RAM, SDD OCZ-ARC100 (240GB), HDD WD Red (3TB), and then some. Headless most of the time, display output to the projector and optionally to a 1920x1200 monitor.
My software: Mageia 8, account (user) "movie" with autologon into KDE, account "music" in a vnc session :7 (starting by systemd at boot), account "browser" in another vnc session :3, account "artee" normally accessed from within the "browser" vnc session (terminal: "su - artee" to get into that account), where I have my private email and browser sessions for banking, all somewhat separated from the regular browsing stuff.
That one specific bit - account artee only being accessed from command line - is very relevant, due to the events that rolled out in short succession after the blasted toaster killed the power....
Anyway, my server plays music in the living room, dims the atmosphere LED lights according to the sunrise and sunset times, and it does a million other things.
Anyway, I've not had to mess with KDE's underpinnings in ages.
And I'm not very aware, since I only know about stuff that I've had to fix (not working as I desired or as it should), and KDE generally works fine.
Now, I had seen that Mageia 8 was out of support since December, I just never got around to upgrading.
The toaster did turn my piece of bread into a nice brown toast, then when sliding down the next slice of bread, our living room went dark. Not fun. My server off.
Turns out, my wife had cleaned the toaster with steel wool, leaving all kinds of steel threads inside, causing a short and thus a power cut.
Strangely enough, the server did start again, but it was beeping on and off.
Well now, as long as I have to connect a monitor to it to check, I might as well upgrade to Mageia 9, right?
So, rebooting, going into the bios, I figured out (after multiple successful boots, the OS was working fine) that the intermittent beeping that became non-intermittent, was actually the fan fail alarm. 0 rpm. But stopping, unplugging, replugging and restarting the system got it to go after a bit, and I changed the speed settings in the BIOS to be less low-noise low-speed. For now, the fan is going (pumping from time to time, so it's on its last legs for sure)...
Okay, now, rebooting into the OS, using the autologin account, I did the cli upgrade dance: removing repos of Mageia 8, pointing to the repos of Mageia 9, running the urpmi upgrade command several times until it didn't have anything to update anymore. Smooth and fast, 10gbit is great (well, only 2.5gbit on the fibre router Eth connection, because somehow my isp are still losers in several ways), never done a faster system upgrade.
Rebooting again into the new kernel, all seems okay. Well, except for the popup message: Harddrive failure is imminent. Not cool. I check what's up, my 4TB HDD is fine, the 250GB SDD is showing loads of very large numbers on smartctl. Since the system is still running, not showing weird behaviour and not crashing, I guess the drive may be problematic for writing, but apparently it's okay for reading. Meanwhile I've gotten a 1TB SATA WD Red SDD as replacement. But dmesg and journalctl show no problems with the current drive, so the replacement can wait. I did copy (rsync from differently mounted root partition to avoid all device-nodes and such, another neat trick that I came across in this journey) over all data (OS, no user data on that SDD.) I also put grub on that new drive, with the proper UUIDs of the new drive to be able to select it instead of the breaking drive. But that's also still to come.
Now the system seems to work, the fan spins (not pumping most of the time), the alarm isn't going, so I figure, let's see if things work.
Regular account movie: video playback works. VNC sessions: music account won't run mpd. Actually, it does start, but can't connect to any internet radio station. It does play mp3 and other local stuff, and one of the internet radio stations does work... Mageia 9 comes with a new version of Curl, which doesn't allow spaces in names, and my mpd playlist of internet radio stations has spaces in the comment with the description after the actual link. This is easily explained, but it took me quite some time to figure that out...
Browser account: the various things I use there seem to work fine.
Account "artee" via su: no kmail.
This is not cool. Retry/investigate: Akonadi errors.
Running Kmail as movie: no problem.
Figuring out "akonadictl start" and such things. No dice.
Akonadi fails to start due to the DB somehow not working.
- Code: Select all
org.kde.pim.akonadiserver: Starting up the Akonadi Server...
org.kde.pim.akonadiserver: database server stopped unexpectedly
org.kde.pim.akonadiserver: Database process exited unexpectedly during initial connection!
org.kde.pim.akonadiserver: executable: "/usr/sbin/mysqld"
org.kde.pim.akonadiserver: arguments: ("--defaults-file=/home/artee/.local/share/akonadi/mysql.conf", "--datadir=/home/artee/.local/share/akonadi/db_data/", "--socket=/tmp/runtime-artee/akonadi/mysql.socket", "--pid-file=/tmp/runtime-artee/akonadi/mysql.pid")
org.kde.pim.akonadiserver: stdout: ""
org.kde.pim.akonadiserver: stderr: ""
org.kde.pim.akonadiserver: exit code: 1
org.kde.pim.akonadiserver: process error: "Unknown error"
org.kde.pim.akonadiserver: Shutting down AkonadiServer...
org.kde.pim.akonadicontrol: Application '/usr/bin/akonadiserver' exited normally...
Time is available in limited slots (family and such). So searches and investigation gets spread over time in my evenings.
More web searches. No solutions. This is more and more starting to look like a real catastrophe. I can send mail through web access (provider inbox is limited to 100MB don't ask me why it's not bigger...). But my emails that I should answer to, and that I want to look stuff up in, etc...?
Lots of searching gets me to learn about KDE and Akonadi, but nothing really similar shows up. I did find quite a few solutions telling to delete some db files, but nothing worked for me (don't delete, just rename; I don't know why people recommend to delete things when people's data is at stake..).
Why does no one have my problem?
Linux desktop users are just a few percent of all PC users, but there are 100s of millions of those.
So that implies a few million users. Mageia users aren't so plentiful, but KDE users should be quite prevalent. Let's say about one 3rd of all desktop users.
Now we're at a million users. Some use Thunderbird, some Evolution, some other Linux mail programs, some Webmail. As a Kmail user, I'm perhaps 1 of several 100 thousand users.
Of those, maybe a few percent use a VNC session. So now I'm one of a few thousand users. But most will shut down neatly when upgrading. I always have. Until this time.
So apparently out of several millions I'm the only one who doesn't use a desktop session for the account with his Kmail, and who didn't shut down neatly before upgrading.
When my PC got the power cut, MariaDB didn't get time to shut down properly. All different instances of MariaDB, of all users. But when I rebooted a couple of times, the various sessions had gotten started and stopped properly. Except the user/account "artee" that I didn't get into to start all applications that I normally run, because why get everything going, if you're going to shut down again shortly for the new kernel and other upgraded software....
Now this hint in the mysql error log in the "artee" account finally gave away what was amiss:
- Code: Select all
[ERROR] InnoDB: Upgrade after a crash is not supported. The redo log was created with MariaDB 10.5.11. You must start up and shut down MariaDB 10.7 or earlier.
Ratz, that sucks!
Now I didn't want to break my current Mageia 9 (repos back to 8, then downgrade MariaDB? I'm adventurous but that sounded like a recipe for disaster), so I decided to install Mageia 8 onto a virtual machine. That took some time (well, fast system, fast internet, it was mostly the configuration that took time), and then I had to figure out which files to move back and forth, then actually do it.
Spoiler, it worked.
And yes, I found out that the mails are in subdirectories in the kmail folder, but I also wanted the configuration and all.
Lessons learnt:
If you don't walk the downtrodden paths, make sure you know what you're doing.
Always shut down neatly before upgrading. Or better, first update everything, then get everything running, then reboot to see everything starting properly, then upgrade.
If you read all this to the end, thanks for reading!!
