Control panel down, websites works

I was working on a website in controlpanel when CP just stopped working. You can’t access is at all, Firefox just says " Unable to connect".

I noted the harddisk is full so I deleted so now there is free space again. I tried to start apnscp but no luck. Restart server a couple of times.

I tested running:

  • /usr/local/apnscp/bin/apnscpd -f
  • upcp -sbf
  • upcp --reset

But after running “systemctl start apnscp” it says:
Job for apnscp.service failed because a timeout was exceeded.
See “systemctl status apnscp.service” and “journalctl -xeu apnscp.service” for details.

Looking at start.log it says:
WARNING: Cache_Mproxy::launch(): Failed to launch Redis server, attempting background repair:
0B. Error_Reporter::add_warning(“Failed to launch Redis server, attempting background repair: %s”, [“”])
[/usr/local/apnscp/lib/log_wrapper.php:95]
1B. warn(“Failed to launch Redis server, attempting background repair: %s”, “”)
[/usr/local/apnscp/lib/Cache/Mproxy.php:83]
2B. Cache_Mproxy::launch(“apnscp”, 600)
[/usr/local/apnscp/lib/constants.php:90]
3B. include(“/usr/local/apnscp/lib/constants.php”)
[/usr/local/apnscp/lib/apnscpcore.php:77]
4B. include(“/usr/local/apnscp/lib/apnscpcore.php”)
[/usr/local/apnscp/lib/lservice.php:25]
WARNING: ListenerService\Daemon::start(): Filesystem virtual root /home/virtual/FILESYSTEMTEMPLATE' uses a different mount point. Duplicating files will double storage usage in this setup. / and /home/virtual/FILESYSTEMTEMPLATE’ should use the same filesystem mount.
[Tue Apr 14 13:55:42 2026] Checking if frontend started

Log looks OK. What’s the result of /usr/local/apnscp/bin/apnscpd -f restart?

Resolved over email.

Annotating for the record:

  • Using Valkey on EL10
  • appendonly.aof.<PORT>.incr.aof ledger was busted in /usr/local/apnscp/storage/run
    • Verified with /usr/sbin/runuser -s /bin/sh apnscp -c cd /tmp && redis-server '/usr/local/apnscp/config/redis.conf' --unixsocket '/usr/local/apnscp/storage/run/redis.sock' --dir '/usr/local/apnscp/storage/run' --pidfile '/usr/local/apnscp/storage/run/redis.pid'
  • Removing the file permitted the panel to successfully resuscitate itself
    • Prior, panel couldn’t set a secure access key from the backend to frontend that resulted in an eventual stack overflow after 30 minutes.