mempool app won't start propperly

[cjvanzwol]

After starting Umbrel, the mempool app won't start. The icon is stuck flashing with the "Starting..." underneath it.

The problem occurred after updating to v0.4.4. The app worked fine in v0.4.3

What i did:

• restart Umbrel
• delete-install from Umbrel webinterface
• stopping-starting from ./scripts/app
• uninstall-install from ./scripts app
• trying to find out if mariadb was the problem:
  1. ./scripts/app stop mempool
  2. rm -r apps/mempool/mysql
  3. ./scripts/app start mempool > the app shows as started in the Umbrel webinterface, but offcourse the app doesn't start propperly
  4. ./scripts/app stop mempool
  5. restored files in apps/mempool
  6. ./scripts/app stop mempool > problem returns

Debug log Umbrel for mempool says:

web_1 | 2021/10/27 08:32:05 [emerg] 7#7: host not found in upstream "mempool.space" in /etc/nginx/conf.d/nginx-mempool.conf:49
web_1 | nginx: [emerg] host not found in upstream "mempool.space" in /etc/nginx/conf.d/nginx-mempool.conf:49

As far as i can see, there is nog such file in de nginx container

[louneskmt]

Hey! You shouldn't have deleted the template mysql dir in apps/mempool. Can you update to v0.4.6 so it can be restored? Then I'll help you fix this 😁

[cjvanzwol]

re-installed mempool again and looked at the debug log again. Got different output:

mempool

mariadb_1 |
mariadb_1 | The problem could be conflicting information in an external
mariadb_1 | my.cnf files. You can ignore these by doing:
mariadb_1 |
mariadb_1 | shell> /usr/bin/mysql_install_db --defaults-file=~/.my.cnf
mariadb_1 |
mariadb_1 | You can also try to start the mysqld daemon with:
mariadb_1 |
mariadb_1 | shell> /usr/sbin/mysqld --skip-grant-tables --general-log &
mariadb_1 |
mariadb_1 | and use the command line tool /usr/bin/mysql
mariadb_1 | to connect to the mysql database and look at the grant tables:
mariadb_1 |
mariadb_1 | shell> /usr/bin/mysql -u root mysql
mariadb_1 | mysql> show tables;

[cjvanzwol]

Hey! You shouldn't have deleted the template mysql dir in apps/mempool. Can you update to v0.4.6 so it can be restored? Then I'll help you fix this 😁

The problem occured even without moving file from the folder. I just used the update button from the Umbrel webinterface.
And after moving them, I restored the template.

But i guess you're saying this stil could be a problem when trying to fix this in my current setup.

I'm currently in the proces of setting up a Ring of Fire, so i'll update to v0.4.6 when that is done, and report back... To Be Continued...

[aphex3k]

I seem to have the same issue as OP initially:

I see this repeating in docker logs -f mempool_web_1:

2021/11/02 23:50:16 [emerg] 7#7: host not found in upstream "mempool.space" in /etc/nginx/conf.d/nginx-mempool.conf:49
nginx: [emerg] host not found in upstream "mempool.space" in /etc/nginx/conf.d/nginx-mempool.conf:49
2021/11/02 23:51:22 [emerg] 8#8: host not found in upstream "mempool.space" in /etc/nginx/conf.d/nginx-mempool.conf:49
nginx: [emerg] host not found in upstream "mempool.space" in /etc/nginx/conf.d/nginx-mempool.conf:49
2021/11/02 23:52:28 [emerg] 7#7: host not found in upstream "mempool.space" in /etc/nginx/conf.d/nginx-mempool.conf:49
nginx: [emerg] host not found in upstream "mempool.space" in /etc/nginx/conf.d/nginx-mempool.conf:49
2021/11/02 23:53:33 [emerg] 7#7: host not found in upstream "mempool.space" in /etc/nginx/conf.d/nginx-mempool.conf:49
nginx: [emerg] host not found in upstream "mempool.space" in /etc/nginx/conf.d/nginx-mempool.conf:49

I'm on latest umbrel 0.4.6 with mempool app docker-compose file from this git repo.

• mempool/frontend:v2.2.2@sha256:515b9a3ab3cc593fe599393b316d4d016335801fbba02ef8c6bb38082597f7cc
• mempool/backend:v2.2.2@sha256:c758345017587dd9e42047528cf6cd71ceae70cebd1a09ce875959259612a2ee
• mariadb:10.5.12@sha256:dfcba5641bdbfd7cbf5b07eeed707e6a3672f46823695a0d3aba2e49bbd9b1dd

[Falci]

Same here.

I was wondering if it could be related to PiHole, but the logs shows mempool.space in green/allowed

[cjvanzwol]

Same here.

I was wondering if it could be related to PiHole, but the logs shows mempool.space in green/allowed

I had PiHole running as well. I wanted to move my PiHole out of umbrel anyway (i want PiHole tot keep running, even if i have tot restart umbrel for any reason), so I uninstalled the PiHole app and then install Mempool again.

Still not working...

[GalvaniObst]

This seems to be a name resolution issue. I had the same error message. After hard coding the DNS server and restarting Umbrel the app mempool worked.

This is what I did:

Edit /etc/docker/daemon.json

{
    "dns": ["1.1.1.1", "8.8.8.8"]
}

Found this here: https://serverfault.com/a/918568
(Thanks wisbucky)

[cjvanzwol]

Tnx @GalvaniObst that solved my problem!

[lukechilds]

Users that are suffering from this, if you visit to mempool.space in your browser on a machine that's on the same network as your Umbrel, does it open properly, or do you get an error?

[GalvaniObst]

When connected via ssh I was able to do a dns lookup and it resolved properly.
So yes, name resolution for mempool.space worked from the raspberry, my machine etc..

As background for my setup.
I'm running pi-hole and my DHCP propagates to use pi-hole for DNS.
Gateway 192.168.0.1
Umbrel 192.168.0.2

I think the container was told to talk to 192.168.0.2 for DNS but wasn't able to reach it.

[lukechilds]

Got it, very helpful thanks @GalvaniObst!

If anyone else can let me know too that would be great.

[gibbs19]

This worked for me too. Thanks a lot!

Note for anyone having the same issue - it's perfectly fine to create file if you don't have a daemon.json already in there. You can either touch the file first or create the file in your editor and save it as a new file.

[cjvanzwol]

Changing DNS solves it.

I changed the docker-compse for the mempool app, and added the dns for the container

services:
  web:
    dns:
        - "1.0.0.1"
        - "1.1.1.1"

But my system keeps revulting my changes after updates of docker of umbrel.

Would be nice to have a permanent solution.

[jonathanalevi]

Still having this problem, despite trying all the suggestions here in dhcpcd and daemon.json. Even Removing Pi-Hole has done nothing at all. So frustrating.

[lukechilds]

Would be nice to have a permanent solution.
@cjvanzwol editing the NDS settings here instead of in the compose files will persist between Umbrel updates.

@jonathanalevi sorry about that, are you able to provide mempool logs so we can take a look? IF you ssh in to your Umbrel you can get the logs with:

~/umbrel/scripts/app compose mempool logs --tail 100

[nmfretz]

closing old support issue