GLPI Inventory Tasks of GLPI Agent

[dethorpe]

Hi all, I hope I am posting this in an appropriate place. I seem to be missing some of the connective tissue between getting the GLPI server and GLPI agent to work together for network and esx inventories.

I believe my setup is 90% there, the GLPI server is running and accepting regular inventory from glpi-agents. glpi-agents have the appropriate modules installed and I can run manual esx inventories by hand (IE #glpi-ex --host xxx.xxx.x.xxx --user foo --password foo2 --directory )

I have created a task on the server with those details above that I "think" should be triggering the agent to do it's thing however that never seems to leave "prepared" and "not done yet" if I am understanding the Job execution information. I also seem to have a blind spot in the documentation here for the interrelationships between the server and agent.

If someone has a minute to help me diagnose, I'd be happy to pay it forward with some time writing up the experience for the documentation.

Thanks a million.

(cross post from glpi-project/doc-agent#4)

[g-bougard]

Hi @dethorpe

first is you want to pay for help, the best way is to earn a GLPI subscription to request professional support. To do so, you can directly contact one of Editor's partners, see: https://glpi-project.org/partners/

Speaking about your question, if your task is not prepared, there can be few reasons for that:

1. you missed to enable the crontab for GLPI and then automatic actions are not started. You should check the journal of automatic actions if they are really started.
2. you missed to enable the related task on the agent configuration in GLPI
3. you missed some configuration on the task
4. you forget to enable the "Permit to re-prepare task after run" on the task
5. you missed to create a VMWare Host credential and to associate it to you "Remote device inventory"

[dethorpe]

about your "Access denied" problem, you should add "127.0.0.1" to the agent httpd-trust parameter.

That did it!

[Thu May 11 10:00:30 2023][info] target server0: server http://xxxx/front/inventory.php

Finally, I see you're using http://xxxx/front/inventory.php URL... It is wrong. Take car of the caution notes here: https://glpi-agent.readthedocs.io/en/latest/configuration.html#server

Thanks for the notes. I corrected that although the address seems to be a default value, as I don't see it explicit anywhere except a comment in agent.cfg
$ grep front /etc/glpi-agent/*
/etc/glpi-agent/agent.cfg:#server = http://server.domain.com/front/inventory.php
Maybe this is worth patching correct in future versions...
I think this is tangental though, I do not think this is causing a problem (especially not the one I am trying to solve) as inventory of the system this agent is installed on reaches GLPI fine but it's ability to inventory remote systems through a task still does not appear to be working.
Where else can we look?
[Fri May 12 12:43:41 2023][info] running task ESX
[Fri May 12 12:43:41 2023][error] [http client] communication error: 404 not found
Could this sequince be indicating an error?

[g-bougard]

I just updated the default config for next nightly to include another sample:
3363642

About you 404 not found... something is probably still wrong in your URL or in your server. You should set debug = 2 in agent config to have a complete debug of what is requested in the agent. For ESX task, this also can be an ESX url issue.

[dethorpe]

At debug level 2, I've had to turn inventory off to just filter down the signal from noise. I hope that isn't a problem. I think the problem is being described here:
[Mon May 15 01:15:33 2023][info] running task ESX
[Mon May 15 01:15:33 2023][debug2] http://DNSNAME/marketplace/glpiinventory/?action=getConfig&task[ESX]=2.7&machineid=GLPI-23-2023-05-04-19-35-07
[Mon May 15 01:15:33 2023][debug2] http://xx.xxx.xxx.xx/glpi/marketplace/glpiinventory/b/esx/?action=getJobs&machineid=GLPI-23-2023-05-04-19-35-07
[Mon May 15 01:15:33 2023][error] [http client] communication error: 404 Not Found
It strikes me as significant that I define the server by hostname in my config yet it appears to have picked up the ip address here for some reason. Either should work but the second has added a to the path in a way that I do not think exhists. Where is it getting this second url? It certain isn't something I have defined.

Me thinks a good reference for someone else someday.
I triggered the job by hand, not just by "force start" but also "restart selected jobs"

[g-bougard]

Now you probably have to fix your "Agent Base URL" on your entity in the "Assets" tab:

If you have an url there, first try to set it empty to leave GLPI Agent to use the default one.

[dethorpe]

We did it!
Thank you.