GIS

Oops... I don't think there's really an option to recover a deleted ticket without digging in the backups of the DB.

For this PGadmin issue in particular, i vaguely remember that there was an operation to do at the end of the setup in order to deal with our environment, so it's a bit annoying.

Will continue from here on this ticket and see how it goes.

I'm in CSR office and could just reboot the GisDB VM: it was stuck at the reboot. Now the Gisaf web services seem to be running correctly.

I am very sorry for the trouble.
Would the e-mails automatically generated by Redmine for each ticket/issue help? I saw that they contain some jargon of what you wrote yesterday...

Giulio

Context¶

Installation of pgadmin4 from:

root@gisdb:/etc/systemd/system# cat /etc/apt/sources.list.d/pgadmin4.list 
deb https://ftp.postgresql.org/pub/pgadmin/pgadmin4/apt/bullseye pgadmin4 main

Packages: pgadmin4-server and pgadmin4-web

Identification of the failure¶

The packages (from the above mentioned origin) cannot work with the current installation.

That includes pgadmin4 in a Python virtual environment - i don't think it was the case with the previous versions.

Moreover, looking at /etc/systemd/system/pgadmin.service, the root of pgadmin4 installation seems to have changed from /usr/share/pgadmin4/web to /usr/pgadmin4/web

The main issue now is then that gunicorn isn't provided in the pgadmin4 virtual environment.

Fix¶

Install the missing package manually:

/usr/pgadmin4/venv/bin/pip install gunicorn

Fix the pgadmin4 service:

root@gisdb:/etc/systemd/system# cat pgadmin.service 
[Unit]
Description=Gunicorn daemon (pgadmin)
Requires=pgadmin.socket
After=network.target

[Service]
Type=notify
# the specific user that our service will run as
User=www-data
Group=www-data
# another option for an even more restricted service is
# DynamicUser=yes
# see http://0pointer.net/blog/dynamic-users-with-systemd.html
RuntimeDirectory=gunicorn
WorkingDirectory=/usr/pgadmin4/web
ExecStart=/usr/pgadmin4/venv/bin/gunicorn --bind=unix:/run/pgadmin.sock --workers=1 --threads=5 pgAdmin4:app
ExecReload=/bin/kill -s HUP $MAINPID
KillMode=mixed
TimeoutStopSec=5
PrivateTmp=true

[Install]
WantedBy=multi-user.target

For the record, here's another important piece:

root@gisdb:/etc/systemd/system# cat pgadmin.socket 
[Unit]
Description=gunicorn socket (pgadmin)

[Socket]
ListenStream=/run/pgadmin.sock
# Our service won't need permissions for the socket, since it
# inherits the file descriptor by socket activation
# only the nginx daemon will need access to the socket
User=www-data
# Optionally restrict the socket permissions even more.
# Mode=600

[Install]
WantedBy=sockets.target

~~~

@Guilio: the interesting bits were in the comments/logs, which aren't copied in the subsequent mails. Hopefully it won't be needed anymore as a quick check shows success after the fixes mentioned above. The thing is that i don't remember exactly the reason for the tweaks (jargon note: maybe something about CSRF token?).

~~~

By the way, the public URL is still: https://gis.auroville.org.in/pgadmin