pgAdmin 4 9.18

[systems368]

Hi Team,
Just upgrade to 9.18 from 9.17 - completley locked out of the system with this error when trying to enter the password to connect to a server.

exception: access violation writing 0x0000000000000000

[systems368]

Completely uninstalled and re-installed (for some reason it held onto the object list for groups and servers). Tried to register a new server - same issue.

[gsleap]

I have the same problem. I also deleted my pgadmin4 and pgadmin folders and it still does this.
Only since pgAdmin 9.18.
Running Windows 11 25H2
p.s. @systems368 might be an idea to have a more descriptive issue title e.g. Access violation writing error after upgrading to 9.18

[gsleap]

This seems to be a dupe of this issue. which has more action

[patcbr]

Confirming this — same error, fresh install of 9.18 today (downloaded directly from the postgresql.org FTP mirror, so not a corrupted/stale download).

A few additional details that might help narrow it down:

The crash isn't specific to password-auth connections. I hit the identical access violation writing 0x0000000000000000 even on a bare trust-auth localhost connection with no password prompt involved at all (Postgres 13, pg_hba.conf rule: host all all 127.0.0.1/32 trust). So this doesn't look tied to the password/keyring flow specifically — it happens on Save/Connect generally.
From the pgAdmin log (pgadmin4.log), the traceback bottomed out here:
File "...\psycopg\connection.py", line 109, in connect
rv = waiting.wait_conn(gen, interval=_WAIT_INTERVAL)
File "...\psycopg\waiting.py", line 118, in wait_conn
fileno, s = gen.send(ready)
File "...\psycopg_connection_base.py", line 441, in _connect_gen
pgconn = yield from generators.connect(conninfo, timeout=timeout)
File "...\psycopg\generators.py", line 84, in _connect
status = conn.connect_poll()
File "...\psycopg\pq\pq_ctypes.py", line 127, in connect_poll
return self._call_int(impl.PQconnectPoll)
File "...\psycopg\pq\pq_ctypes.py", line 814, in _call_int
return func(self._pgconn_ptr)
OSError: exception: access violation writing 0x0000000000000000

So it's faulting specifically inside the ctypes call into libpq.dll at PQconnectPoll, not in pgAdmin's own Python logic.

Ruled out a local environment issue on my end: no conflicting libpq.dll/Postgres tooling anywhere on PATH, clean uninstall + wipe of both %APPDATA%\pgAdmin and %APPDATA%\pgAdmin4 before the 9.18 install, reproducible from a totally fresh state.
Downgrading to 9.17 with identical connection settings works perfectly — no crash, connects instantly. So this looks like a genuine regression introduced between 9.17 and 9.18, likely in the bundled libpq/psycopg3 build for Windows rather than anything on the pgAdmin Python side.

Environment: Windows 11, build 10.0.26200.9457, pgAdmin 9.18 x64 desktop install, connecting to Postgres 13.

[rsprinkle]

Also confirming the the same error:

Version 9.18
Application Mode Desktop
Commit f09776c 2026-09-15
Python Version 3.13.14
Current User pgadmin4@pgadmin.org
Electron Version 44.3.0
Browser Chrome 152.0.7977.78
Operating System Windows-11-10.0.26100-SP0

[dev-hari-prasad]

Duplicate of #10428.

For anyone facing this error until a patch is released, you can fix this by:

1. In the Object Explorer, right-click your server and select Properties (or configure it when registering a new server).
2. Go to the Parameters tab.
3. Click the + button to add a new parameter:
4. Name: GSS encmode (or gssencmode)
5. Value: disable
6. Click Save and reconnect.

Or by downgrading to 9.17.

Please discuss this in issue #10428 if you need to, and avoid using this issue; as rightly pointed out by @gsleap, there is more material in #10428.

[abanoub-fathy]

@dev-hari-prasad you are a live saver