Official Discourse Advertising plugin results in complete server crash after install (Breaks my instance after I install it)

[ServerMan3932]

Hey, I've been checking out discourse and wanted to monetize it with ads. I found the discourse advertising plugin and it had all the features I needed to start configuring advertisements, it's also officially supported so I thought it should be good to go.

I installed it from the GitHub repo linked on their Official Meta Instance using the instructions on the Cloudron Docs and then when I restarted the discourse app; it started back up completely broken (I make backups before I install plugins though, so I was able to restore thankfully)

Upon the server restarting, this is the message I get:

Oops
The software powering this discussion forum encountered an unexpected problem. We apologize for the inconvenience.
Detailed information about the error was logged, and an automatic notification generated. We'll take a look at it.
No further action is necessary. However, if the error condition persists, you can provide additional detail, including steps to reproduce the error, by posting a discussion topic in the site's feedback category.

The site is completely down when this message displays, so ironically, the feedback category is down with the site.

My process consisted of:

• 1. I opened a web terminal for Discourse and did: cd /app/code/plugins
• 2. I did: git clone https://github.com/discourse/discourse-adplugin.git
• 3. I shut down the Discourse app, and turned it back on again
• 4. The website gives out and displays the error.

It's been a headache trying to get working so any help would be appreciated!

[girish]

Is there any error in the logs?

Also, do you face this issue just with this plugin or other plugins as well?

[ServerMan3932]

@girish I've faced this issue with a few different plugins unfortunately. I just gave up on trying to run the other quality-of-life type plugins, but the ad plugin is one I really need

Here's a list of the things that I've tried and failed at running so far:

• (Officially supported) The Discourse Ad Plugin. (Crashes server on startup)

• (Officially supported) The Discourse Reactions Plugin (Installs and shows up as disabled but crashes server with that error message upon enabling, rebooting doesn't help either, it just starts back up with the error)

• (Third Party Plugin) Discourse Custom Wizard (Crashes server on startup, same result as the ad-plugin)

• (Officially Supported) Spoiler Alert Plugin (Crashes server on startup)

Some errors I get on both a normal boot (No ad plugin) and a bad boot (ad plugin installed) are:

fatal: not a git repository (or any of the parent directories): .git

CRIT Supervisor is running as root. Privileges were not dropped because no user is specified in the config file. If you intend to run as root, you can set user=root in the config file to avoid this message.

Not sure why the above issues would be a problem to this specifically, but they may be worth mentioning

As far as a comparison between when the plugin is installed vs when it's not, the difference comes at the very end as NGINX is starting up.

On a bad boot with the Ad-Plugin Installed; the logs look like this,

2023-09-11T14:40:00.000Z ==> Starting discourse
2023-09-11T14:40:00.000Z => Healtheck error: Error: connect ECONNREFUSED :80
2023-09-11T14:40:01.000Z 2023-09-11 14:40:01,334 INFO spawned: 'unicorn' with pid 174
2023-09-11T14:40:01.000Z 2023-09-11 14:40:01,335 INFO spawned: 'nginx' with pid 175
2023-09-11T14:40:02.000Z 2023-09-11 14:40:02,883 INFO success: nginx entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2023-09-11T14:40:02.000Z 2023-09-11 14:40:02,883 INFO success: unicorn entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)

On a good boot it's this,

2023-09-11T14:36:24.000Z ==> Starting discourse
2023-09-11T14:36:25.000Z 2023-09-11 14:36:25,654 INFO spawned: 'unicorn' with pid 718
2023-09-11T14:36:25.000Z 2023-09-11 14:36:25,655 INFO spawned: 'nginx' with pid 719
2023-09-11T14:36:27.000Z 2023-09-11 14:36:27,212 INFO success: unicorn entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2023-09-11T14:36:27.000Z 2023-09-11 14:36:27,213 INFO success: nginx entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)

I also checked my browser java-script console during the error, its completely empty when the error displays.

I've also compared the installed version to the supported version,

The Ad-Plugin Github Repo states the supported version is 1.4. The latest Cloudron package is 3.1.0.

Nothing seems off there because it's officially supported and was just recently updated, so I would imagine that means anything above 1.4 too.

The last thing I can think of is that it could be trying to write to a protected area, because one of the biggest things done differently from the official package is that cloudron makes things immutable, so maybe something has to be symlinked? (Could be totally left field on that, just spitballing)

I think there's likely a broader issue with plugins on the cloudron package of discourse that may need some attention, because official plugins shouldn't be doing anything too weird like a third party/unsupported plugin might; so my biggest guess is that these plugins expect something that's being denied to them.

[nebulon]

Those logs all seem fine actually. That one healthcheck failure just happens if the healthcheck hits a timeframe during app startup. You mention a few times "crashes server on startup" what is that crash and do you have any logs for this?

[ServerMan3932]

@nebulon Hey, I took another look at the logs and found that right before it configures compiles the CSS; it's deleting and renaming things related to the plugin. Not sure if that's normal or not but It's worth mentioning.

execute("DELETE FROM site_settings\nWHERE name IN (\n 'codefund_property_id',\n 'codefund_advertiser_label',\n 'codefund_advertiser_short_label',\n 'codefund_through_trust_level',\n 'codefund_nth_post',\n 'codefund_display_advertiser_labels',\n 'codefund_below_post_enabled',\n 'codefund_above_post_stream_enabled',\n 'codefund_above_suggested_enabled',\n 'codefund_top_of_topic_list_enabled'\n)\n")
2023-09-11T14:39:32.000Z -- execute("UPDATE site_settings SET name = 'ads_txt' WHERE name = 'adsense_ads_txt'")
2023-09-11T14:39:32.000Z -> 0.0009s
2023-09-11T14:39:32.000Z -> 0.0012s
2023-09-11T14:39:32.000Z == 20190603112536 RenameSiteSettingAdsTxt: migrated (0.0030s) =================
2023-09-11T14:39:32.000Z == 20190603112536 RenameSiteSettingAdsTxt: migrating ==========================
2023-09-11T14:39:32.000Z == 20200703082449 RemoveCodefundSiteSettings: migrated (0.0016s) ==============
2023-09-11T14:39:32.000Z == 20200703082449 RemoveCodefundSiteSettings: migrating =======================
2023-09-11T14:39:33.000Z ==> Setup OpenID connect integration
2023-09-11T14:39:34.000Z fatal: not a git repository (or any of the parent directories): .git
2023-09-11T14:39:38.000Z fatal: not a git repository (or any of the parent directories): .git
2023-09-11T14:39:40.000Z => Healtheck error: Error: connect ECONNREFUSED :80
2023-09-11T14:39:40.000Z fatal: not a git repository (or any of the parent directories): .git
2023-09-11T14:39:44.000Z fatal: not a git repository (or any of the parent directories): .git
2023-09-11T14:39:47.000Z fatal: not a git repository (or any of the parent directories): .git
2023-09-11T14:39:48.000Z ==> Skip building assets (already built)
2023-09-11T14:39:50.000Z => Healtheck error: Error: connect ECONNREFUSED :80
2023-09-11T14:39:50.000Z fatal: not a git repository (or any of the parent directories): .git
2023-09-11T14:39:51.000Z -------------
2023-09-11T14:39:51.000Z Compiling CSS for default 2023-09-11 14:39:51 UTC
2023-09-11T14:39:51.000Z Start compiling CSS: 2023-09-11 14:39:51 UTC

As far as crashing goes, I meant that the website ceases all functionality and displays this message to users.

Oops
The software powering this discussion forum encountered an unexpected problem. We apologize for the inconvenience.
Detailed information about the error was logged, and an automatic notification generated. We'll take a look at it.
No further action is necessary. However, if the error condition persists, you can provide additional detail, including steps to reproduce the error, by posting a discussion topic in the site's feedback category.

[ServerMan3932]

@ServerMan3932 I fixed it! I ended up fixing it with a server-upgrade. Turns out, the crashing was caused by insufficient resources to run all the plugins, but it's working better now.

[nebulon]

Ah glad you found the root cause for this. Were there not any out-of-memory notifications in Cloudron which might have hinted at this or were those somehow below Cloudron's radar?

[ServerMan3932]

@nebulon I didn't see any warning signs of low memory, but it just started working after I switched from a shared CPU to a dedicated one.