[Elastic Agent] - RPM / DEB not working, Agent can't enroll to fleet - shows message: "Usage: /etc/init.d/elastic-agent {start|stop|status|restart|force-reload}"

[EricDavisX]

On enrolling the .deb 7.11-SNAPSHOT package agent, we are getting the following message
"Usage: /etc/init.d/elastic-agent {start|stop|status|restart|force-reload}"
and are unable to enroll the .deb agent using the following steps on Linux Debian 9 with .deb or a Centos8 with .rpm instance (verified on both, but deb example is given below:

Steps followed:

Download the "elastic-agent-7.11.0-amd64.deb" package artifact page.
Agent: https://snapshots.elastic.co/7.11.0-2a00e604/downloads/beats/elastic-agent/elastic-agent-7.11.0-SNAPSHOT-amd64.deb

place the agent package on home/zeus location on Debian 9 Linux machine.

Run the command to get the root/admin permission: sudo su root

sudo dpkg -i elastic-agent-7.11.0-SNAPSHOT-amd64.deb

sudo elastic-agent enroll https://df36b1e2a07749b8842cecf42ae7358a.us-central1.gcp.foundit.no:9243 SG5ZdS1IVUI2UFEzTEg4aEgyWGI6VEUzbnB6QjRRYWF0aW1Bbkg5XzBCUQ==

Observe "Usage: /etc/init.d/elastic-agent {start|stop|status|restart|force-reload}" message is displayed.

Execute sudo /etc/init.d/elastic-agent start
and observe that .deb agent is not displayed in UI.

screenshot

[elasticmachine]

Pinging @elastic/ingest-management (Team:Ingest Management)

[EricDavisX]

thank you @dikshachauhan-qasource @rahulgupta-qasource for bringing this issue, I logged it so it could be tracked more immediately and we can clean up the other test issue where it was first noted, #22549 (comment)

[EricDavisX]

if anything is changing / changed in our intended usage from 7.10 to 7.11 please do let us know and we can track updates to the Docs too, this page: https://www.elastic.co/guide/en/fleet/current/elastic-agent-installation.html

[EricDavisX]

I verified the same on a known working good Centos8 image - the .rpm install shows the same as .deb example.

[blakerouse]

This looks to be a machine configuration issue and not with elastic-agent itself.

I think the issue is your machine as /etc/init.d in the PATH environment variable. That should never be in your PATH. Remove /etc/init.d from your PATH environment variable then the /usr/bin/elastic-agent should be picked up.

[EricDavisX]

I'm seeing a separate error on a RHEL7 image:

[zeus@mainqa-atlcolo-10-0-7-174 elastic-agent-7.10.0-linux-x86_64]$ sudo ./elastic-agent install -f --kibana-url=https://asdf.eastus2.azure.elastic-cloud.com:443 --enrollment-token=asdf==
The Elastic Agent is currently in BETA and should not be used in production

Error: failed to install service (elastic-agent): exit status 1

And I can confirm a separate, different RHEL7 image works. : /

Note: we got the .rpm / .deb tests added back in to e2e-testing, here:
elastic/e2e-testing#439
... and there, I'm seeing a failure. Tho the code and centos/debian (public) images were working prior to the install feature refactor. It may yet be that all of them have /etc/init.d in PATH. We can cross that off the list and open individual tickets if needed, thanks for the report in Blake. I'll reply back with more details.

[EricDavisX]

oh I'm sorry - I got confused again on what I was tracing. The above note is mis-leading... I was testing on RHEL7 with .tar.gz package and found a case that didn't work and shouldn't have tied that to this issue.
...I'll log the above RHEL7 issue separately. : /

To clarify:
I checked the $PATH on the centos docker container that the e2e-testing system uses and it didn't have .init.d in the $PATH at all. So, it seems a separate issue from this / that problem. See screenshot:

I also tested the Centos8 image I said I had tested, adn find the same problem exists as tested in 7.10.1:

@blakerouse do you want a separate issue for the e2e-testing container scenario? Maybe not until we can figure out the issue? I am glad that we keep finding things and fixing them for .deb/.rpm it is hard going tho. Maybe using the e2e-testing repo test (that is currently marked and being merged as 'skipped') will allow a super quick way to test and iterate over this.

[blakerouse]

@EricDavisX can you provide the output of which elastic-agent?

[EricDavisX]

@blakerouse sure:
[zeus@mainqa-atlcolo-10-0-7-165 ~]$ which elastic-agent
/usr/bin/elastic-agent

[blakerouse]

@EricDavisX Then what is the output of cat /usr/bin/elastic-agent?

[EricDavisX]

looks like codez. does this look right to you for version 10.0.1:
elastic-agent-file.txt

[blakerouse]

Okay something is wrong with the RPM package then. That is the file that should be placed in /etc/init.d and a smaller simple wrapper should be installed in /usr/bin/elastic-agent.

I never ran into this issue in my testing, something must have changed in the package.yml that broke this.

[ph]

@blakerouse Did you investige it?

[blakerouse]

@ph Yes, seems the package building is copying the wrong file to the /usr/bin/elastic-agent file. Still investigating why, because the packages.yml looks correct.

[blakerouse]

I think I figured it out. Happens that I made the files the same name, so when the package is build the /etc/init.d script overwrites the /usr/bin script. Then when the package is installed the overwritten file place it both paths.

I have a fix I am testing now, should be a simple fix, if I am correct.

[ghost]

Hi @EricDavisX

We have performed regression testing for Linux deb/rpm on 7.10.2 BC1 Kibana cloud build. Build details is as follows:

Platform: Staging
Version: 7.10.2
Commit: a0b7936
Build number: 36136
Artifact: https://staging.elastic.co/7.10.2-e6b2b8c2/summary-7.10.2.html

We have execute test cases under test plan execution :

• Test Run for Linux .deb deploy of BC Agent/Endpoint with minimal sanity test
• Test Run for Linux .rpm deploy of BC Agent/Endpoint with minimal sanity test

Please let us know if anything else is required to be covered.

Thanks