I know this is a forum for LE, but I think some of the moderators frequent the Kodi forums. Anyway, my account appears to be closed on the Kodi forum. I can login, but then I can't browse or do anything (including contact a forum admin). I think this is because many years ago I requested the account to be closed for personal reasons at that time. Who can I contact to help me unlock or reopen my Kodi forums account? Thanks.
Circling back to this. I am still on MariaDB 10.5.1 and have since upgraded LE to 9.2.6 and other Kodi versions to 18.9 without any issues. I have not attempted to upgrade MariaDB since then.
And I found this about MariaDB affecting Kodi databases. I have yet to try the solutions, but it looks promising. Yes, I know that is CoreELEC. The state of MariaDB in the context of CoreELEC - Kodi - CoreELEC Forums
brand new MariaDB 10.5.4 container with no previous databases existing. reinstalled LE 9.2.4 on a single RPi3. Reinstalled Kodi 18.8 on a single Mac. Launched Kodi on the Mac and created the new databases. Scanned in a folder of media. Powered on the RPi3 (http://ix.io/2ysv). Closed then opened Kodi on the Mac and it just crashes (https://pastebin.com/raw/H4jtv7G7). Rebooted the LE machine and it still works, but Kodi on the Mac crashes.
Dropped all databases. Installed Kodi 18.7 on the Mac and downgraded LE to 9.2.3 (found an image backup I had). Powered on the RPi3 to create the new databases and scanned in a folder of media. Launched Kodi on the Mac and cannot play any media (https://pastebin.com/raw/vubS8VUA). Rebooted LE machine and database is fine and can play media (http://ix.io/2ysG). Reopened Kodi on the Mac and database partially shows up but won't play any media.
MariaDB logs for all of this: https://pastebin.com/raw/TbY6jCBJ
So on a whim, I decided to spin up a fresh MariaDB 10.5.1 container. LE 9.2.3, Kodi 18.7. Brand new databases, scanned in a folder of media. No issues. Rebooted both LE and the Mac and relaunched Kodi a number of times. No issues. I can select and play media on both. Restore my old database and still no issues on either. Upgrade LE to 9.2.4 and database shows up fine and I can play media. Update Kodi on Mac to 18.8 and no issues. Mac Kodi 18.8 log https://pastebin.com/raw/n1h6HmDC, LE 9.2.4 log http://ix.io/2ysT
I'm now on the latest LE and Kodi release, but on a slightly older MariaDB release but the database is working again cross platform. I tried the same steps as above with MariaDB 10.5.2 and I get the database corruption showing "0" and Mac Kodi just crashes. I also tried doing an inline upgrade of MariaDB from 10.5.1 to 10.5.2 with the same crashing results.
Yes, wait for network is enabled with the default timeout of 10 seconds on all LE machines. The reason it shows unknown database is because I had dropped the databases and was letting Kodi generate them from scratch (per the instructions).
the aborted connection warnings show up in the MariaDB log whenever I close a MAC Kodi player or reboot/shutdown an LE machine. I run the MariaDB 10.5.5 official docker container on a Synology NAS. The database is not shutting down or restarting. I have one other application that writes to a completely separate database under a different user and it has not experienced any read/write errors or performance issues. Before this Kodi database issue, the MariaDB docker container had an up time of over 40 days.
the mysqld shutdown is me actually restarting the database container because I could not drop the databases after I did my testing for some reason (that was the first time it had ever done that, it got hung dropping)
just a reminder, this all worked under LE 9.2.3 and Kodi 18.7. MariaDB was upgraded from 10.5.4 to 10.5.5 at some point, but I could not tell you when that was. I believe it was prior to updating LE or Kodi. There was a part of that upgrade that seemed to fail but was fixed by running mysql-upgrade again. If I can recall, it was an issue with the sql_mode column in the mysql.proc table not of the correct type. After running the upgrade again, that error has not come up again.
Here's my steps:
- shutdown all Kodi players
- Drop Kodi databases in MariaDB
- start 1 LE machine and let it generate the database (http://ix.io/2yov)
- shutdown LE machine
- restore database in MariaDB
- start 1 LE machine and database shows up (http://ix.io/2yoz)
- start 1 Mac Kodi player and it just hangs and will not actually start (https://pastebin.com/raw/9j5cb7Cs)
- restart 1 LE machine and database shows up (http://ix.io/2yoD)
Then I did the opposite startup
- shutdown all Kodi players
- Drop Kodi databases in MariaDB
- start 1 Mac Kodi player and let it generate the database (https://pastebin.com/raw/NzMFmsM0)
- shutdown Mac Kodi player
- restore database in MariaDB
- start 1 Mac Kodi player (https://pastebin.com/raw/xRewtmSQ)
- start 1 LE machine and database is corrupted and shows "0" for info (http://ix.io/2yoH)
- reopen 1 Mac Kodi player and database shows up but cannot play any media and crashes (https://pastebin.com/raw/M0Y3Fu02)
I run MariaDB in a docker container on the NAS. I believe the log is piped directly to the docker container log. I'm trying to to export that (Synology GUI doesn't play well with docker log, so I have to find the actual file it's reading).
EDIT: And here is the only MariaDB log file I could find. https://pastebin.com/raw/dcamzQzA
yes I should've added the full log. Sorry. Here is the log from the Mac after an LE machine has connected to the database and Kodi hangs on start and crashes.
Do you want me to grab a log from the LE machine after I restore the database and start it for the first time?
Here is the LE Kodi log from the LE machine I was testing with.
Yes, I saw the release notes about the fix for MariaDB on Android, but I assumed that was specific to Android and should not have affected my two types of players (Mac and LE).
I have the following setup:
- media hosted on a NAS
- MariaDB 10.5.5 on NAS hosting MyVideos116 and MyMusic72 databases
- 4 x LibreELEC 9.2.4 on an RPi3
- 3 x Macs running Kodi 18.8
When the first RPi3 starts it modifies the database somehow that causes the 3 Macs running Kodi to not be able to start and just crash. The logs from the Mac states something about the database and then just crashes (or hangs).Code
If I shutdown all LE players and then restore the database, the Mac Kodi players work just fine. It isn't until I turn on or restart one of my RPi3 running LE 9.2.4 that the database corrupts and the Mac players start crashing due to the database. Additionally, when this corruption happens, all of the media on the main screen of the LE players has the title of "0". The progress bars for in progress media is correct and even the watched totals are correct, but thumbnails, posters, titles, info, etc are all replaced with "0" on screen and I cannot select or play anything. Additionally, I cannot update the database and scan for media until I drop and restore the database. If I do not start up any of my Mac Kodi players, then all of the LE machines work off of the database it creates.
So each player type works when using the database it creates/uses. But the other player type cannot use that database. This all worked fine under LE 9.2.3 and Kodi 18.7. Unfortunately the LE add on does not allow downgrading from 9.2.4 to 9.2.3. This database issue also happens with LE 9.2.4 (Kodi 18.8) and Kodi 18.7 on the Macs, which I find interesting because theoretically point releases should all work off the same database, only the major releases introduce breaking database changes as shown here: Databases - Official Kodi Wiki
Terribly sorry for the long delay. I've recently moved and just got everything set back up.
I changed the squash settings to "No Mapping" and it looks like I can SSH in and it accepts the password and appears to be working. Kudos to mglae for helping me out. looks like the squash settings was the last piece to making this work. Now if I can just get the booting to be a little more reliable (about 1 in 4 boots fails at a black screen and no activity light)... but that's for another thread and may not be LibreELEC related.
Unfortunately I don't use Ubuntu. My setup is a Synology NAS hosting the NFS share, Unifi using DHCP Network Boot options to point clients to the NAS TFTP/PXE server hosting bootcode.bin. This is the guide I followed for the NFS share: Network Booting LibreElec to Raspberry Pi's from a Synology NAS - Upods.netUpods.net and this link for the TFTP/PXE server: DiskStation Manager - Knowledge Base | Synology Inc. and I think I found a few pointers for the Unifi configuration on their forums but don't remember exactly where I found them.
From the first guide, the squash settings are set to map all users to admin. I followed all other settings in that guide. I might have changed 1 or two things like the exact folder path, but none of the other settings changed that I remember. Here's a screenshot of my Synology NFS share permissions:
The options in the squash settings dropdown are:
- No Mapping
- Map root to admin
- Map root to guest
- Map all users to admin
- Map all users to guest
my /etc/exports file does not look like what is in that wiki guide:
And with this being on a Synology, I'm not quite sure what would happen manually editing that file. It may get overwritten, or not survive reboots, etc.
Even on a fresh install the files/folders get owned by "admin". This most likely has to do with the pxeboot setup, but I'm not sure. That's the user "1024" in place of the root user on the ls -al command. If I try to chown -R root /storage I've experienced both the inability to boot and an error in the command line effectively saying it wouldn't let me.
Apologies for the delay in response. Thank you mglae for working through this with me.
systemctl status sshd returns:Code
I'm assuming you meant add textmode and not "fextmode". I have added it again.
Still looks like I have a permissions issue in my initial setup of the pxeboot because if I try to ssh out of the PXE client even to localhost, it gives me an error:Code
EDIT: I deleted all of the files in the pxeboot storage location and started from scratch. SSH starts, and I am in textmode but when I try to ssh to localhost it asks for the password (default at this time), I type it in and it just says Permission denied, please try again. So my permission issues have been resolved, but now I'm back to the default password not working.
systemctl status ssh afterwards reports 3 failed password attempts for root and a connection closed.
Deleted the files in the storage folder for the RPi on the TFTP server and set all folders to have the appropriate owner. Restarted the RPi and everything booted up as expected and this time it actually ran me through the setup wizard. I changed my settings accordingly and attempted to SSH in however it's not accepting the default SSH password. I changed the password using the LibreELEC settings add-on, restarted the SSH service and restarted the RPi but it still won't let me SSH. I can get it to prompt me for a password, but it fails saying incorrect password.
systemctl status ssh
reports: Unit ssh.service could not be found.
journalctl -u sshd
reports the sshd.service exited with code "FAILURE" and appears the reason is it could not save the public keys and therefore could not find any public keys to start with. The specific error can be found in the log posted below.
Looks like maybe I have some permissions issues on the TFTP server for PXE booting... which is odd because I can access, save, and modify files from another machine.
ok I added `textmode` to the command line file and got a shell on boot. Of course when I did this I failed to realize I needed a keyboard plugged into the pi... This is where my knowledge of Linux/LibreELEC and the underlying system starts to dwindle. What logs would I be looking for? I'm assuming I need to begin with making sure the SSH service is started first.
On a side note, I just updated to LE 9.2 and am successfully booting over pxe, however I'm back to square one with the connection refused when trying to SSH from another machine and the enable SSH option is missing from the LE add-on.
I successfully have an RPi3 booting LibreELEC 9.2 over PXE. Everything in that respect works. It’s really just for proof of concept to test everything out.
The issue I am having is SSH no longer works. My commandline.txt file has quiet ssh ip=dhcp at the end of it. In Kodi under the LibreELEC settings Add-On at first the option to enable SSH was not there under the Services tab, so I removed ssh from my commandline.txt and rebooted and now it shows up and shows as on, regardless if I keep it removed or put it back in commandline.txt
However, when I open a terminal from another computer on the same network and type ssh [email protected] (that’s the correct IP) I get ssh: connect to host 192.168.1.30 port 22: Connection refused. I tried a fresh install again and dropped a blank file titled “ssh” in the boot folder but that didn’t change anything.
What am I doing wrong or what do I need to do to get SSH working? Is it something to do with booting over PXE and technically no local install, rather I’m theoretically trying to ssh into the server over NFS instead?
My issue doesn't appear related to TVH. I used Kodi about 2-2.5 hours prior to the crash. The TV and AVR the Pi is connected to was on at the time of the crash, but not set to the Kodi input. I only knew it rebooted because I have LEDs behind my TV controlled by Boblight and they turned on at the restart. I was not doing anything on my NAS nor my network at the time. UniFi doesn't report any network interruptions and it shows everything up and online. No Docker containers were restarted either manually or automatically. Here's some more logs from the Kodi crash about 10 minutes ago. I really don't know what I'm looking at or for in the logs. Also the output of dmesg | paste again.
That’s an interesting find. I have experienced situations where while watching a channel through the TVH plugin it freezes video but I still have audio for about a minute then it stops all together and I have to cycle power or issue a reboot command over SSH. But this is rare, though it could be related. Now that I think about it, the times it has rebooted are somewhat synchronized but not exactly in line with when I have updated the Docker container for TVH on my server. I’d have to do some more testing. Can anyone show me in the logs I posted anything that might lead to making this a valid conclusion?