Dpinger lots of cpu

dennypage

The individual command lines for dpinger look fine, but there clearly is a problem in that you have duplicate copies of dpinger running for most of the gateways.

I recall there was a problem with early beta snapshots that resulted in duplicate dpinger instances. Are you running a very old snapshot?

Ofloo

I reinstalled today to do these tests with an image i've downloaded yetserday and since this morning i've updated twice.

dennypage

I've reached out to the gent that handles the dpinger start/stop code and asked him to have a look.

Ofloo

Let me know if you there's anything I can do to help, .. or if I can provide some output.

cmb

Could you send me a backup of your config? PM here or email cmb at pfsense dot org. The copy from status.php (which removes password hashes, certificates, etc.) would suffice.

Ofloo

done

Ofloo

amd64
10.3-RELEASE
FreeBSD 10.3-RELEASE #7 51f8df0(RELENG_2_3): Mon Apr  4 08:10:48 CDT 2016     root@ce23-amd64-builder:/builder/pfsense/tmp/obj/builder/pfsense/tmp/FreeBSD-src/sys/pfSense

Crash report details:

PHP Errors:
[04-Apr-2016 20:55:58 Europe/Brussels] PHP Stack trace:
[04-Apr-2016 20:55:58 Europe/Brussels] PHP   1\. {main}() /etc/rc.filter_configure_sync:0
[04-Apr-2016 20:55:58 Europe/Brussels] PHP   2\. filter_configure_sync() /etc/rc.filter_configure_sync:37
[04-Apr-2016 20:55:58 Europe/Brussels] PHP   3\. filter_generate_gateways() /etc/inc/filter.inc:288
[04-Apr-2016 20:55:58 Europe/Brussels] PHP   4\. return_gateway_groups_array() /etc/inc/filter.inc:881
[04-Apr-2016 20:55:58 Europe/Brussels] PHP   5\. return_gateways_status() /etc/inc/gwlb.inc:891
[04-Apr-2016 20:55:58 Europe/Brussels] PHP   6\. get_dpinger_status() /etc/inc/gwlb.inc:421
[04-Apr-2016 20:55:58 Europe/Brussels] PHP   7\. stream_socket_client() /etc/inc/gwlb.inc:346

Ofloo

Strange I changed my dns servers in general settings, had to do it twice because for some reason like gateways it won't take it after the first time, .. but now dpinger has settled, .. with max of 10% which seems very acceptable to me ? Not sure how this relates, .. the widget of openvpn is doing something strange not updating, .. says all the vpns are down while there not.

I'll reboot tomorrow and see if it stays that way.

edit:

i also removed few firewall rules, .. if it stays this way then i'll email you the updated configuration, .. so you can compare.

screenshot.png_thumb

Ofloo

only unbound and route tend to go high, .. unbound sometimes 69% and route 20% ? seems high but anyways.

sceenshot.png_thumb

Harvy66

99.6% idle and 100% load. Yeah, "load" is a loaded term in Unix. It includes IO wait among other things as part of the calculations.

Ofloo

Not sure what it is you mean.

The 99% idle load you're referring to is probably because it's going up and down and hasn't been updated. There's an update delay on the processes. Just my 2c.

Harvy66

Depending on your flavor of Unix, load includes some percentage IO wait time. Even if a process is not using any CPU, a process can dramatically contribute to the "load" of the system if it's blocking on an IO event. I'm not sure what all blocking IO events contribute to load. Would be strange to include network IO.

But like you mentioned, it's probably something simpler, like update timing.

Ofloo

I'm not sure why dns matters when it comes to dpinger, but its got something to do with it.

Do not use the DNS Forwarder as a DNS server for the firewall

When I check the checkbox, the cpu remains at 0-15%, however when I uncheck the checkbox it shoots to 150% and stays there.

Ofloo

here's something weird, .. so i started to look into this dns thing, and what i found out was it was because the dns server didnd't work that dpinger settled down, .. i recently changed the first dns server to 172.16.1.10 which apparently didn't respond to everything except the firewall, probably some gateway issue, however but since it didn't respond, cpu usage went down!?