Bug 2063930

Confirming the fix, as described in patch http://pkgs.devel.redhat.com/cgit/rpms/irqbalance/commit/?h=rhel-9-main&id=3c91db6799b96664c6b402522562ebbe67fa27eb

# rpm -q irqbalance
irqbalance-1.8.0-5.el9.x86_64

# less /usr/lib/systemd/system/irqbalance.service | grep 'CPUs'
ConditionCPUs=>1

So far, so good; but when I tried on a virtual machine with a single CPU with the _old_ irqbalance, it did not start, saying this:

# rpm -q irqbalance
irqbalance-1.8.0-4.el9.x86_64

# systemctl start irqbalance
# systemctl status irqbalance
○ irqbalance.service - irqbalance daemon
     Loaded: loaded (/usr/lib/systemd/system/irqbalance.service; enabled; vendor preset: enabled)
     Active: inactive (dead) since Fri 2022-05-27 15:30:01 CEST; 7s ago
   Duration: 22ms
       Docs: man:irqbalance(1)
             https://github.com/Irqbalance/irqbalance
    Process: 17741 ExecStart=/usr/sbin/irqbalance --foreground $IRQBALANCE_ARGS (code=exited, status=0/SUCCESS)
   Main PID: 17741 (code=exited, status=0/SUCCESS)
        CPU: 12ms

May 27 15:30:01 kvm-01-guest01.lab.eng.brq.redhat.com systemd[1]: Started irqbalance daemon.
May 27 15:30:01 kvm-01-guest01.lab.eng.brq.redhat.com /usr/sbin/irqbalance[17741]: Balancing is ineffective on systems with a single cpu.  Shutting down
May 27 15:30:01 kvm-01-guest01.lab.eng.brq.redhat.com systemd[1]: irqbalance.service: Deactivated successfully.

So, it seems it has some kind of awareness of this problem by itself; aren't we double-patching a single problem here?

Otherwise, I can confirm that the patch works, and with it, systemd does not even try to load irqbalance if there is only a single CPU, as planned:

# rpm -q irqbalance
irqbalance-1.8.0-5.el9.x86_64

# systemctl start irqbalance
# systemctl status irqbalance
May 27 15:50:13 kvm-01-guest01.lab.eng.brq.redhat.com systemd[1]: irqbalance daemon was skipped because of a failed condition check (ConditionCPUs=>1).

So everything is OK and I am ready to stamp VERIFIED when confirmed that we want to doubly handle this situation.

(In reply to Jiri Dluhos from comment #2)
> Confirming the fix, as described in patch
> http://pkgs.devel.redhat.com/cgit/rpms/irqbalance/commit/?h=rhel-9-
> main&id=3c91db6799b96664c6b402522562ebbe67fa27eb
> 
> # rpm -q irqbalance
> irqbalance-1.8.0-5.el9.x86_64
> 
> # less /usr/lib/systemd/system/irqbalance.service | grep 'CPUs'
> ConditionCPUs=>1
> 
> So far, so good; but when I tried on a virtual machine with a single CPU
> with the _old_ irqbalance, it did not start, saying this:
> 
> # rpm -q irqbalance
> irqbalance-1.8.0-4.el9.x86_64
> 
> # systemctl start irqbalance
> # systemctl status irqbalance
> ○ irqbalance.service - irqbalance daemon
>      Loaded: loaded (/usr/lib/systemd/system/irqbalance.service; enabled;
> vendor preset: enabled)
>      Active: inactive (dead) since Fri 2022-05-27 15:30:01 CEST; 7s ago
>    Duration: 22ms
>        Docs: man:irqbalance(1)
>              https://github.com/Irqbalance/irqbalance
>     Process: 17741 ExecStart=/usr/sbin/irqbalance --foreground
> $IRQBALANCE_ARGS (code=exited, status=0/SUCCESS)
>    Main PID: 17741 (code=exited, status=0/SUCCESS)
>         CPU: 12ms
> 
> May 27 15:30:01 kvm-01-guest01.lab.eng.brq.redhat.com systemd[1]: Started
> irqbalance daemon.
> May 27 15:30:01 kvm-01-guest01.lab.eng.brq.redhat.com
> /usr/sbin/irqbalance[17741]: Balancing is ineffective on systems with a
> single cpu.  Shutting down
> May 27 15:30:01 kvm-01-guest01.lab.eng.brq.redhat.com systemd[1]:
> irqbalance.service: Deactivated successfully.
> 
> So, it seems it has some kind of awareness of this problem by itself; aren't
> we double-patching a single problem here?

Hi Jiri,

Thanks for reporting this, I think it is the same issue addressed in upstream [1].
Irqbalance do has the check in it's main function, which will fail due to one cpu, but it doesn't stop systemd from trying to start the service, which is a waste of effort. So this patch simply stop systemd from starting the service when cpu_num=1.

Thanks,
Tao Liu


[1]: https://github.com/Irqbalance/irqbalance/pull/201

(In reply to Jiri Dluhos from comment #3)
> Otherwise, I can confirm that the patch works, and with it, systemd does not
> even try to load irqbalance if there is only a single CPU, as planned:
> 
> # rpm -q irqbalance
> irqbalance-1.8.0-5.el9.x86_64
> 
> # systemctl start irqbalance
> # systemctl status irqbalance
> May 27 15:50:13 kvm-01-guest01.lab.eng.brq.redhat.com systemd[1]: irqbalance
> daemon was skipped because of a failed condition check (ConditionCPUs=>1).
> 
> So everything is OK and I am ready to stamp VERIFIED when confirmed that we
> want to doubly handle this situation.

Yes, looks good to me.

Thanks,
Tao Liu

Agreed; setting VERIFIED.

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory (irqbalance bug fix and enhancement update), and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHBA-2022:8328