CentOS 8 probs [Solved]

[crimson30]

EDIT:
See attachment to this post for my personal instructions that worked in the end!
Note: I was unable to get this working with media in /home and had to make sure / had enough space for 4TB of media (not the default)
The rest of this thread saved for posterity.


I'm trying to get UMS working on CentOS 8 with no luck. I made a CentOS 7 VM to make sure the instructions I had handy were correct and it worked just fine on a fresh CentOS 7 install. Also, I used

Code: Select all

sudo update-alternatives --config java

to switch to 11, since that's what I have running on the VM.

I'm not seeing a debug.log file in the ums folder

Code: Select all

[crimson30@ivy ums]$ sudo systemctl status ums
● ums.service - Universal Media Server
   Loaded: loaded (/etc/systemd/system/ums.service; enabled; vendor preset: disabled)
   Active: activating (auto-restart) (Result: exit-code) since Tue 2020-03-10 13:18:26 MDT; 7s ago
  Process: 5056 ExecStart=/opt/ums/UMS.sh (code=exited, status=203/EXEC)
 Main PID: 5056 (code=exited, status=203/EXEC)
[crimson30@ivy ums]$ sudo journalctl _SYSTEMD_UNIT=ums.service
-- Logs begin at Tue 2020-03-10 12:23:36 MDT, end at Tue 2020-03-10 13:18:43 MDT. --
-- No entries --
[crimson30@ivy ums]$ ls
BUILD.md       documentation   DummyInput.jpg  LICENSE.txt  logback.headless.xml  plugins    renderers  ums.jar  VirtualFolders.conf  WEB.conf
CHANGELOG.txt  DummyInput.ass  INSTALL.txt     linux        logback.xml           README.md  UMS.conf   UMS.sh   web
[crimson30@ivy ums]$

The only changes I made to the attached .conf are the following:

Code: Select all

network_interface = enp5s0
folders = ~/Music
folders_monitored = ~/Music

Little help?

[crimson30]

I changed UMS.conf to point toward /home/crimson30/Music and:
1. It seems to be ignoring the folders and folders_monitored parameters and trying to look at files in /home/crimson30/archive. What's up with that?
2. Although I have "network_interface = enp5s0" in there, there's this in the trace: INFO 16:04:45.635 [main] Using address /192.168.122.1 found on network interface: name:virbr0 (virbr0)

I've attached some command line output (as well as the slightly updated UMS.conf).

update: I ran

Code: Select all

systemctl disable libvirtd.service

... and now that virbr0 is gone, it finds the right interface. Still, why does it ignore UMS.conf?

I get the feeling it's not even looking at /opt/ums/UMS.conf (operating directory) or /etc/ums/UMS.conf

[crimson30]

When I run

Code: Select all

strace /opt/ums/UMS.sh /etc/ums/UMS.conf

, the media server is visible to other computers, but the folders are incorrect.

If I run the service, it doesn't work.

By the way, this is what I'm using that has worked in the past for /etc/systemd/system/ums.service:

Code: Select all

[Unit]
Description=Universal Media Server
 
[Service]
Type=simple
Restart=always
RestartSec=90
Environment="UMS_PROFILE=/etc/ums/UMS.conf"
User=ums
Group=ums
ExecStart=/opt/ums/UMS.sh
 
[Install]
WantedBy=multi-user.target

[crimson30]

Some headway: I tried running /opt/ums/UMS.sh and got more verbose output showing that it couldn't write to /home/crimson30/.config/UMS/UMS.conf

After chmod on /home/crimson30/.config/UMS/UMS.conf, it's working by directly running ./UMS.sh but still not getting the directory right.

Any ideas on how to fix the folder problem or getting it to run as a service?


I'll work on it more tomorrow.

[boss]

I'm running Gentoo linux.
The UMS.conf file resides in /user directory/.config/UMS/UMS.conf. Which in your case would be: /home/crimson30/.config/UMS/UMS.conf
That is assuming that crimson30 is the user account which is starting UMS.

These folder permission work for me (which in your case would be under /home/crimson30/.config/
drwxr-xr-x crimson30 crimson30 UMS
-rw-r--r-- crimson30 crimson30 UMS/UMS.conf
-rw-r--r-- crimson30 crimson30 UMS/UMS.cred
-rw-r--r-- crimson30 crimson30 UMS/WEB.conf

drwxr-xr-x crimson30 crimson30 UMS/data
-rw-r--r-- crimson30 crimson30 UMS/data/UMS.tmpmgr
-rw-r--r-- crimson30 crimson30 UMS/data/UMSKeys.db
-rw-r--r-- crimson30 crimson30 UMS/data/pms.pid

drwxr-xr-x crimson30 crimson30 UMS/database
-rw-r--r-- crimson30 crimson30 UMS/database/medias.mv.db
-rw-r--r-- crimson30 crimson30 UMS/database

Also in your startup script, the line:
Environment="UMS_PROFILE=/etc/ums/UMS.conf"
Maybe should be:
Environment="UMS_PROFILE=/home/crimson30/.config/UMS/UMS.conf"

[Nadahar]

It sounds clear to me too, that like Boss says, the configuration file in use isn't the once you're editing. UMS will use a different profile folder (configuration files are found in the "profile folder") for different users by default. When running as a daemon, the profile folder will depend on which user the daemon is running as. You can however specify the profile folder path using command line parameters, which makes it possible to override the default behavior. If you fail to do that, the default, user dependent, location will apply.

If you use a command line argument to override the profile folder, make sure that the user running the service has sufficient permissions to the folder in question.

[crimson30]

I did find that adding lines to ~/.config/UMS/UMS.conf did effect UMS.conf changes, however, by making changes to selinux, I could instead edit /etc/ums/UMS.conf

I figured something out: CentOS 8 seems to have changed something about its selinux setup (another possibility is that it's something to do with a change to the media location from /archive/~library4-mov to ~/archive/~library4-mov because I believe there are some changes to the CentOS server install wizard which led to me doing so). When running

Code: Select all

sudo journalctl -xe

, I saw selinux errors.

I fixed things a bit by doing the following:

Code: Select all

su
ausearch -c '(UMS.sh)' --raw | audit2allow -M my-UMSsh
semodule -i my-UMSsh.pp
/sbin/restorecon -v /opt/ums-9.3.0/UMS.sh

And *now* when I edit /etc/ums/UMS.conf, it reflects the changes. And *now* the service doesn't fail/exit.

I just need to figure out how to keep selinux from denying access to the media folder. Interestingly enough, if I run:

Code: Select all

strace /opt/ums/UMS.sh

..it's bypassing the problem and everything works perfectly fine. But when I run the service, the service starts, but it can't access the folder:
"Mar 11 12:12:32 ivy UMS.sh[3162]: WARN 12:12:32.666 [main] "/home/crimson30/archive/~library4-mov" does not exist. Please remove it...."

I'm no selinux wiz, so I just need to dig around a little to get it working. I don't want to disable selinux.

(Note: in previous examples, I was using ~/Music for simplicity, but the actual media location is to be ~/archive/~library4-mov)

[Nadahar]

I always disable selinux as the very first thing I do unless the computer is going to be facing the internet (web server etc), so I have nothing to contribute in that regard. I guess I figure life is too short...

[crimson30]

I temp set selinux to permissive with

Code: Select all

setenforce Permissive

and it's still giving me that

Code: Select all

WARN 12:12:32.666 [main] "/home/crimson30/archive/~library4-mov" does not exist. Please remove it....

I've tried chmod -R 777 /home/crimson30/archive/~library4-mov

...and I still get that problem. And it still works fine with strace or directly running the script in /opt/ums (./UMS.sh).

[crimson30]

Strange. It doesn't like when media is under /home.

I moved everything to /archive instead of /home/crimson30/archive and it's working.

And yes, I had everything fully pathed in UMS.conf (I just used ~ as shorthand when posting).

Anyway, it would seem that most of my problems stemmed from putting all the media files under /home.

The only reason I put the files under /home this time was because I was getting frustrated with the OS install wizard and didn't make a separate partition for data like I usually do. I think they made a change to the disk management part. And then I ended up sticking everything under /home because / only had <50GB of space and /home had 4TB.