Solved How to disable Solr

[infogulch]

I enabled Solr to try it out, but using 3GB is a bit much on my 4GB server and a bunch of my apps started crashing with OOM errors, so I need to disable it. I didn't see the option to disable it in mail settings, so I lowered the mail memory down to 512MB, which (perhaps predictably) caused even more problems and now the box app is not responding at all.

I'm able to ssh into the server, but neither the cloudron web app nor cli works and instead gives "Login failed: Status code: 500 message: connect ECONNREFUSED 127.0.0.1:3306".

Any ideas to un-f*** myself?

[BrutalBirdie]

@infogulch hello.

First of all what do you mean by I enabled Solr to try it out?
If you installed solar on the root system then just uninstall it..

[nebulon]

Does it not work if you disable solr again by unchecking the same checkbox where you enabled it?

[infogulch]

@nebulon There was no checkbox. The email config page said that solr was enabled, but the button to open the config dialog was just missing.

This button was just absent:

@BrutalBirdie I'm referring to the Full Text Search feature in Email settings which is powered by Solr.

[girish]

@infogulch Clicking that button is supposed to bring up the dialog again which will let you disable solr. Do you see any errors in the browser console when you click that button?

[infogulch]

@girish I'm saying the button was missing entirely. There was no button.

the button to open the config dialog was just missing.

Edit: to be clear, the screenshot above came directly from the cloudron.io docs pages, sorry if that was confusing.

And I can't get to the admin page right now because it's been down since this occurred.

[girish]

@infogulch Oh ok, so the first step is to get the admin page up. I would stop the mail container first to get back some memory, docker stop mail. Then systemctl restart box. Does this get the dashboard up? If not, can you check /home/yellowtent/platformdata/logs/box.log ?

[infogulch]

mail container is stopped, after systemctl restart box the logs transitioned from

2021-05-10T18:04:40.001Z box:apphealthmonitor run: could not check app health. connect ECONNREFUSED 127.0.0.1:3306
Box GET /api/v1/cloudron/status 500 Internal Server Error connect ECONNREFUSED 127.0.0.1:3306 1.351 ms - 217
Box GET /api/v1/cloudron/status 500 Internal Server Error connect ECONNREFUSED 127.0.0.1:3306 1.692 ms - 217
Box GET /api/v1/cloudron/status 500 Internal Server Error connect ECONNREFUSED 127.0.0.1:3306 2.145 ms - 217
Box GET /api/v1/cloudron/status 500 Internal Server Error connect ECONNREFUSED 127.0.0.1:3306 2.381 ms - 217
2021-05-10T18:04:50.002Z box:apphealthmonitor run: could not check app health. connect ECONNREFUSED 127.0.0.1:3306
Box GET /api/v1/cloudron/status 500 Internal Server Error connect ECONNREFUSED 127.0.0.1:3306 1.853 ms - 217

... into this:

2021-05-10T18:04:51.732Z box:box Received SIGTERM. Shutting down.
2021-05-10T18:04:51.733Z box:tasks stopTask: stopping all tasks
2021-05-10T18:04:51.734Z box:shell stopTask spawn: /usr/bin/sudo -S /home/yellowtent/box/src/scripts/stoptask.sh all
2021-05-10T18:04:56.425Z box:server ==========================================
2021-05-10T18:04:57.946Z box:server ==========================================
2021-05-10T18:04:59.324Z box:server ==========================================
2021-05-10T18:05:00.540Z box:server ==========================================
2021-05-10T18:05:02.024Z box:server ==========================================

Which if I tail the log it is still printing every second or so.

However the my. admin page is still returning a 503.

[girish]

@infogulch It seems the box code is crashing instantly on start up. Can you check if there is disk space (df -h) and also if mysql is running using systemctl status mysql.

[infogulch]

@girish df -h shows 23G Avail, so I suppose it's not a full disk.

systemctl status mysql showed

● mysql.service - MySQL Community Server
     Loaded: loaded (/lib/systemd/system/mysql.service; enabled; vendor preset: enabled)
     Active: inactive (dead) since Sun 2021-05-09 20:01:10 UTC; 22h ago

systemctl start mysql brought the admin page back up.

[infogulch]

And after starting the mail container, I was able to disable solr.

Everything seems to be back online again. Thanks for your help.

My guess is that some general slowness (perhaps due to memory overcommit and swap?) was preventing the mail app from responding to queries about its status, which I percieved as "the button didn't show up".

[girish]

@infogulch said in How to disable Solr:

My guess is that some general slowness (perhaps due to memory overcommit and swap?) was preventing the mail app from responding to queries about its status, which I percieved as "the button didn't show up".

Yes, exactly this. I can actually reproduce this. If the mail container is not responding, the button is not appearing. Maybe I will show some spinner or equivalent instead of the UI disappearing with no feedback.

[infogulch]

@girish I think that would be a good change.

I also think I should have gone through the Troubleshooting document sooner :

https://docs.cloudron.io/troubleshooting/

Thanks!