My PostgreSQL 15 database is completely stuck in a crash-recovery loop. The database ran out of disk space during a migration and now cannot complete recovery because it needs to write WAL files but has no disk space remaining.

What happened:

1. I was upgrading Umami analytics from v2.9.0 to v3.0.3

1. A migration tried to UPDATE ~2 million rows which filled the disk

1. PostgreSQL crashed with "No space left on device"

1. Now it cannot complete recovery because recovery itself needs disk space

Error from logs:

2025-12-14 12:27:38.836 UTC [29] FATAL:  could not write to file "pg_wal/xlogtemp.29": No space left on device

2025-12-14 12:27:38.843 UTC [2] LOG:  startup process (PID 29) exited with exit code 1

2025-12-14 12:27:38.843 UTC [2] LOG:  shutting down due to startup process failure

Recovery loop evidence:

The database keeps trying to recover but fails every ~30 seconds:

2025-12-14 12:25:53.472 UTC [29] LOG:  redo starts at 19/E40CB370

2025-12-14 12:25:59.830 UTC [29] LOG:  redo done at 1A/19FFE950 system usage: CPU: user: 2.91 s, system: 1.97 s, elapsed: 6.35 s

2025-12-14 12:25:59.863 UTC [29] FATAL:  could not write to file "pg_wal/xlogtemp.29": No space left on device

Important:

• The redo completes successfully (redo done at 1A/19FFE950)

• Recovery fails only when trying to write the checkpoint

• Data appears intact - just needs disk space to complete recovery

Database size:

Approximately 3.7 GB based on earlier successful query

What I need:

1. Increase the PostgreSQL volume size (I have Pro plan with 100GB allocation but can't find volume resize option)

1. OR give me shell access to manually delete old WAL files to free space

Plan:

Railway Pro (100GB disk allocation available)