Since 2.4.4 random latency spikes

[mrsunfire]

Hi!

Since upgrade from 2.4.3 to 2.4.4 I‘ve noticed some random latency spikes on both WAN interfaces. The latency on ALL gateways goes up to 100ms, even the LAN_DHCP gateway.

Also ping graph from think broadband shows this issue:

2.4.3

2.4.4

The other WAN is the same. Any hints?

[jimp]

Is this something you can predict?

Maybe sit watching top -aSH and try to see what is happening at the time of the spike.

[mrsunfire]

@jimp said in Since 2.4.4 random latency spikes:

top -aSH

This could be hard because it's only a few seconds (5-10s). It feels like the system hangs, but the GUI is working flawless.

[jimp]

Anything in any of your logs around the time of the spikes? The main system log, gateway log, and DHCP log especially.

[mrsunfire]

Yes. Systemlog:

Oct 4 04:14:53	php-fpm	31026	/rc.filter_configure_sync: Keep current gateway, its already part of the group members.
Oct 4 04:14:53	php-fpm	31026	/rc.filter_configure_sync: Keep current gateway, its already part of the group members.
Oct 4 04:14:52	php-fpm	77540	/rc.filter_configure_sync: Keep current gateway, its already part of the group members.
Oct 4 04:14:52	check_reload_status		Reloading filter
Oct 4 04:14:52	php-fpm	77540	/rc.filter_configure_sync: Keep current gateway, its already part of the group members.
Oct 4 04:14:52	php-fpm	42796	/rc.newwanip: The command '/sbin/route delete -host ' returned exit code '64', the output was 'route: destination parameter required route: usage: route [-46dnqtv] command [[modifiers] args]'
Oct 4 04:14:52	php-fpm	42796	/rc.newwanip: The command '/sbin/route delete -host ' returned exit code '64', the output was 'route: destination parameter required route: usage: route [-46dnqtv] command [[modifiers] args]'
Oct 4 04:14:52	php-fpm	42796	/rc.newwanip: The command '/sbin/route delete -host ' returned exit code '64', the output was 'route: destination parameter required route: usage: route [-46dnqtv] command [[modifiers] args]'
Oct 4 04:14:52	php-fpm	42796	/rc.newwanip: rc.newwanip: on (IP address: 10.1.1.1) (interface: SMARTHOME[opt2]) (real interface: igb1).
Oct 4 04:14:52	php-fpm	42796	/rc.newwanip: rc.newwanip: Info: starting on igb1.
Oct 4 04:14:51	php-fpm	74692	/rc.filter_configure_sync: Keep current gateway, its already part of the group members.
Oct 4 04:14:51	php-fpm	74692	/rc.filter_configure_sync: Keep current gateway, its already part of the group members.

DHCP:

Oct 4 04:14:29	dhclient		Creating resolv.conf
Oct 4 04:14:29	dhclient		RENEW
Oct 4 04:14:29	dhclient	20831	DHCPACK from 10.124.224.1
Oct 4 04:14:29	dhclient	20831	DHCPREQUEST on igb0 to 255.255.255.255 port 67

Nothing in gateway logs.

[jimp]

DHCP renew on the interface could explain the appearance of a momentary glitch

[mrsunfire]

Might be possible! I have the problem that my Philips Hue bridge is crashing if Avahi is beeing used. This could explain that this glitch gets triggered.
I have to setup the bridge to the same interface as my LAN that it works. Unfortunately Avahi only works 80% and let the bridge crash.

[jimp]

There is a rewrite of Avahi coming, it's on 2.4.5 snapshots right now for testing. It might help.

[mrsunfire]

Maybe I give it a try. Can you recommand the snapshot release right now for beeing "stable"?

[jimp]

The latest one is generally the most stable but they are still snapshots. We don't test them all individually before they go up, so there could be problems there.