Large Alias Causing CPU spikes and ping latency

[franco]

We were testing with this regarding the kernel panic issue:

# opnsense-patch https://github.com/opnsense/core/commit/c8497ac14603

It needs a cron apply or reboot.  It will only reload the aliases twice per hours at the expense of the update interval. It has downsides, but we were discussion making it configurable for some edge cases like this.


Cheers,
Franco

[guenti_r]

Hi franco,

it works! Many thanks! Please make the reload configurable in future.

[CanIKipThis]

Thanks how do you do a cron apply after the install?  I have installed the patch yet the cron for that job still is configured for every hour.

[guenti_r]

System -> Settings -> Cron -> "apply"

[CanIKipThis]

Thanks! Did you see a difference for the update.py job in cron after the patch was installed?

[guenti_r]

Yes, the minute latency spikes are gone.

[jphylips]

But I guess they now occur every half hour?

[franco]

That would be the working theory. ;)

[CanIKipThis]

That I get, but I'm just wondering where the logic changed because my crontab still says every minute:



This was after installing the patch and then rebooting.  Did the same thing again and did the crontab apply.

[guenti_r]

No, you are wrong.

Code Select Expand

1,31    *       *       *       *       (/usr/local/bin/flock -n -E 0 -o /tmp/filter_update_tables.lock /usr/local/opnsense/scripts/filter/update_tables.py) > /dev/null

Use crontab -e  and not the GUI

[CanIKipThis]

Thanks thats where I was checking from.  I had installed it at least two times before.  Just did it again, followed cron apply and now it is showing the change.

Thanks for following up!

[jphylips]

Hi Franco,
Thanks for the patch. Great work (as always, and I mean that in a very positive way!!!).
Would it be possible to make the execution moments of the cron job customizable?

[guenti_r]

@franco

is this patch included in 25.1.4 now?

[jphylips]

Judging by this output, I don't think so:

root@OPNsense:~ # crontab -l|grep update_tables
*   *   *   *   *   (/usr/local/bin/flock -n -E 0 -o /tmp/filter_update_tables.lock /usr/local/opnsense/scripts/filter/update_tables.py) > /dev/null
root@OPNsense:~ #

[franco]

No, because it breaks MAC table updates by taking up to 30 minutes to render these changes. We're likely going to make this configurable, but it's not an immediate concern. Just change the values on your end for now.

What ships in 25.1.4 is the performance improvement of the task previously discussed here.


Cheers,
Franco