rcu_preempt detected stalls on CPUs/tasks

Questions / Help
1

On around 12/30/2025, it seems like these log lines appear regularly in one of our regular service’s fly logs:

$ fly logs -a my-app
...
2025-12-30T20:24:15Z app[xxx] ewr [info][xxx] rcu: INFO: rcu_preempt detected stalls on CPUs/tasks:
2025-12-30T20:24:15Z app[xxx] ewr [info][xxx] rcu:  0-...0: (5 GPs behind) idle=ffd4/0/0x1 softirq=221/222 fqs=873082
2025-12-30T20:24:15Z app[xxx] ewr [info][xxx] rcu:  3-...0: (9 ticks this GP) idle=b404/0/0x1 softirq=206/208 fqs=873082
2025-12-30T20:24:15Z app[xxx] ewr [info][xxx] rcu:  (detected by 1, t=4396086 jiffies, g=-395, q=42284 ncpus=4)
...

As far as I’m aware, this service hasn’t had its functionality altered in any way that could cause this to appear; I’m guessing there’s a chance that it’s an issue with the specific host running this vm?

2

I’ve not seen it, but things that may be worth considering:

  • Your machine is subject to CPU throttling; have a look in the CPU graphs in Grafana to see if you’re hitting into the throttle line
  • Your machine became suspended and then woke up again, and this is the weirdness of the internal clocks becoming unsynchronised
1 Like
3

the rcu_preempt is an issue on our end that sometimes (rarely) happens for some unknown reason. try restarting/recreating the machine?

1 Like
4

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.