Issue installing SentinelAgent local rpm with rpm-ostree #482
Comments
|
Can you post the content of the RPM package scriptlets? You can get them via: |
travier
added
external
|
Sure thing, here you go @travier 3K lines of scriplet |
|
I just want to note here that this is a 3k line long scriptlet (which in general is not a good idea) and thus this might be hard to debug. |
|
Yes i totally understand :) |
|
Looks like the error is thrown here: Do you already have a |
|
@LorbusChris How would i go about cleaning out that file in /sysroot as that is read-only? Thanks EDIT: I ran And i now have the file on the fs again: |
|
There are errors, but does the installation actually fail? Finding the unit in Edit: My bad, it's from the repo, not a deployment, so it did fail, and it is expected that the unit file is in the repo (as cache). |
|
Oh, this is running in |
|
I think so yes, this is the output from the command |
|
There isn't much we can do here. They need to fix their pre-installation scriptlet. |
|
I see, thank you. Is this error specific to Fedora Sericea (ostree) or any .rpm based distro? Thanks |
|
This is specific to rpm-ostree based distros (currently Silverblue, Kinoite, Sericea, Onyx, CoreOS, IoT). |
|
Closing this since there's really not that much we can do. Thanks for the assistance! |
|
I worked around this with Run To install: You may or may not need to import the SentinelOne GPG signing key in order to work around an |
|
This RPM would also be a good candidate to convert to a sysext: https://github.com/travier/fedora-sysexts |
|
Oh interesting, this is the first I've heard of sysexts. I'll have to read more about them. Are you thinking (proprietary) upstream would package this as a sysext? |
|
For those following along at home, SentinelOne attempts to keep mutable state in sudo mkdir /var/sentinelone/
sudo cp -a /usr/lib/opt/sentinelone/configuration /var/sentinelone
sudo cp -a /usr/lib/opt/sentinelone/log /var/sentinelone/log
sudo cp -a /usr/lib/opt/sentinelone/crash_dumps /var/sentinelone/
echo '/var/sentinelone/configuration /usr/lib/opt/sentinelone/configuration none defaults,bind 0 0' | sudo tee -a /etc/fstab
echo '/var/sentinelone/log /usr/lib/opt/sentinelone/log none defaults,bind 0 0' | sudo tee -a /etc/fstab
echo '/var/sentinelone/crash_dumps /usr/lib/opt/sentinelone/crash_dumps none defaults,bind 0 0' | sudo tee -a /etc/fstab
sudo systemctl daemon-reload # Not sure if this is strictly necessary, but `mount` warned me about it
sudo mount /usr/lib/opt/sentinelone/configuration
sudo mount /usr/lib/opt/sentinelone/log
sudo mount /usr/lib/opt/sentinelone/crash_dumpsTurns out you'll need to manually enable the SentinelOne systemd service. You can do this with |
It's unlikely upstream would do that, but you can try it using the examples from the repo I linked above. |
Agreed, I guess I was mostly asking if the intended use case is for third party packaging or if the system is designed around upstreams providing these (which would mean that it's good to know about personally, but probably not going to help this particular problem) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Describe the bug
Installing a local .rpm package from SentinelOne results in prein scripts failing with the following message:
Output of journalctl -t 'rpm-ostree(SentinelAgent.prein)':
To Reproduce
Download signed SentinelOne Agent RPM file and install it via
rpm-ostree install SentingelAgent.rpmExpected behavior
I expect the package to be installed in the same manner as if i were running Fedora Workstation 38.
Screenshots
If applicable, add screenshots to help explain your problem.
OS version:
Additional context
I originally created this issue in coreos repo: coreos/fedora-coreos-tracker#1481
The text was updated successfully, but these errors were encountered: