Problems with Discourse install in VMWare

(RBoy) #1

I’m continuing a thread from here:

https://meta.discourse.org/t/how-to-change-the-logo-url/25543/22?source_topic_id=62986

If I goto /admin/upgrade I see:

Repository Name Status
discourse (f4cc71d)
New Version Available!
Remote Version: (13e489b)
Last Updated: 5 hours ago
2 new commits
Upgrade to the Latest Version
discourse-details Up to date
discourse-nginx-performance-report Up to date
docker_manager (146e2de) Up to date
lazyYT Up to date
poll Up to date

It gives me a link, but if go back to dashboard page it says I’m upto date, is this normal or a bug?

How do you update Discourse?
(RBoy) #2

Something isn’t right, it’s been an hour and it’s still upgrading:

discourse (f4cc71d)
New Version Available!
Remote Version: (13e489b)
Last Updated: 6 hours ago
0 new commits

It says Currently Upgrading and when I click on it I get a blank console with Upgrading greyed out

(RBoy) #3

hah so much for stable, I rebooted the machine and now discourse won’t even start (i.e. website won’t connect). Regret trying to upgrade it. Anyone have any suggestions on where to start debugging

(Rafael dos Santos Silva) #4

Can you please try rebuilding?

Please SSH into your server and run:

cd /var/discourse
git pull
./launcher rebuild app
(RBoy) #5

When I do a git pull

Already up-to-date.

When I do a launcher rebuild

./launcher rebuild app
Cannot connect to the docker daemon - verify it is running and you have access

(Jeff Atwood) #6

Docker is no longer working on your system, you should try to fix that.

(RBoy) #7

Yes I got that but no clue how, I love how the whole setup was dummy proof, now I need to understand how all the daemons and stuff works :slight_smile:
Any suggestions on a guide or something or where to start or should I just trash the whole thing and reinstall (I just installed it last night to dip my feet)

(RBoy) #8

So I googled and tried to restart docker, it complained about some PID file still existing in /var/run, deleted it and then tried to start it:

sudo dockerd
INFO[0001] libcontainerd: previous instance of containerd still alive (1604)
WARN[0002] failed to rename /var/lib/docker/tmp for background deletion: %!s(). Deleting synchronously
INFO[0002] [graphdriver] using prior storage driver: aufs
Error starting daemon: error while opening volume store metadata database: timeout

Sigh…here goes the weekend - and all I did was click on update and reboot!

(Jeff Atwood) #9

Sounds like your host isn’t really compatible with Docker. We recommend Digital Ocean. Try installing there.

(RBoy) #10

? It was working fine till I did an update to discourse for that wizard setup bug. Ubuntu 16.04 LTS running in a VM.

(Jeff Atwood) #11

Your VM is incompatible with Docker.

(RBoy) #12

You mean VMWare is incompatible or VM’s are incompatible? I don’t know much about docker but it sounds a little funny to me that ALL of Ubuntu can run and graphics can run but Docker (as software) is hardware dependent.

(Jeff Atwood) #13

No idea, but we don’t see these kinds of problems on Ubuntu on metal.

(RBoy) #14

Oddly it went belly up when I forced a reboot because Discourse was stuck at upgrading that patch put in a few hours ago. Okay lets start over and figure it out.

(RBoy) #15

Do you use get.docker.com or Choose Docker CE or Docker EE on Ubuntu | Docker Documentation for your metal deployments?

Your GitHub recommends using get.docker.com (discourse/INSTALL-cloud.md at master · discourse/discourse · GitHub) but according to Dockers website one should not use that anymore (it installs docker-engine instead of docker-ce).

(Jeff Atwood) #16

I dunno, what do you think there @sam? Should we change our install instructions?

(Sam Saffron) #17

Its all compatible, what is happening here is that he is running launcher as end user which has no access to docker…

try

cd /var/discourse
sudo ./launcher rebuild app

Discourse docker works fine under VMWare, I know cause I run it under VMWare

(RBoy) #18

yep am running it under sudo and I get this

sudo dockerd
INFO[0001] libcontainerd: previous instance of containerd still alive (1604)
WARN[0002] failed to rename /var/lib/docker/tmp for background deletion: %!s(). Deleting synchronously
INFO[0002] [graphdriver] using prior storage driver: aufs
Error starting daemon: error while opening volume store metadata database: timeout

It was working fine until I decided to click on upgrade discourse through the admin/upgrade page and then it just “hung”, I rebooted the machine and all it stopped working (see above)

(Sam Saffron) #19

why are you trying to run dockerd by hand? it is already running, either systemd or whatever init system you have would have started it.

does sudo docker ps return anything?

(RBoy) #20

I did that to debug the problem, see from this post onwards:

After I clicked upgrade (to bring in a patch made by @techAPJ) it just hung for more than an hour, so I rebooted the VM and the next thing I see is that I can’t connect to the server. Then @Falco suggests that I do a ./launcher rebuild app so I decided to run that and I get:

Cannot connect to the docker daemon - verify it is running and you have access

So the next thing I go about trying to figure out is what’s going on with docker and I googled and I found that I can get the status using sudo dockerd and that’s how we ended up here (all because I just had to click on upgrade - when will I learn!) :wink: