[Beta] Asuswrt-Merlin 384.14 Beta is now available

[jsarcone]

I would consider flashing the Beta 1 again and RTFD after. Start from scratch, eliminate the possibilities.

There is something definitely wrong. I flashed it again and after about a few hours and depending on the load; it loses connection to the Internet. I waited little longer and was able to login to Web Interface. I was able to grab logs and CPU 2 was pigged. Attempted to restart the WAN interface and it did work.

logs:

Nov 16 14:11:20 crond[643]: time disparity of 807245 minutes detected

Nov 16 20:55:06 nat: apply redirect rules

Nov 16 20:55:06 WAN_Connection: ISP's DHCP did not function properly.

Nov 16 21:02:33 rc_service: httpd 645:notify_rc restart_wan_if 0

Nov 16 21:02:34 kernel: Attempt to kill tasklet from interrupt

Nov 16 21:02:34 kernel: et0: et_mvlan_netdev_event: event 9 for vlan1 mvlan_en 0

Nov 16 21:02:34 kernel: et0: et_mvlan_netdev_event: event 2 for vlan1 mvlan_en 0

Nov 16 21:02:34 kernel: et0: et_mvlan_netdev_event: event 13 for vlan1 mvlan_en 0

Nov 16 21:02:34 kernel: et0: et_mvlan_netdev_event: event 1 for vlan1 mvlan_en 0

Nov 16 21:02:34 kernel: et0: et_mvlan_netdev_event: event 13 for vlan2 mvlan_en 0

Nov 16 21:02:34 kernel: et0: et_mvlan_netdev_event: event 1 for vlan2 mvlan_en 0

Nov 16 21:02:34 kernel: et0: et_mvlan_netdev_event: event 4 for vlan1 mvlan_en 0

Nov 16 21:02:41 rc_service: httpd 645:notify_rc restart_wan_if 0

Nov 16 21:02:41 kernel: Attempt to kill tasklet from interrupt

Nov 16 21:02:41 kernel: et0: et_mvlan_netdev_event: event 9 for vlan1 mvlan_en 0

Nov 16 21:02:41 kernel: et0: et_mvlan_netdev_event: event 2 for vlan1 mvlan_en 0

Nov 16 21:02:41 kernel: et0: et_mvlan_netdev_event: event 9 for vlan2 mvlan_en 0

Nov 16 21:02:41 kernel: et0: et_mvlan_netdev_event: event 2 for vlan2 mvlan_en 0

Nov 16 21:02:41 kernel: et0: et_mvlan_netdev_event: event 13 for vlan1 mvlan_en 0

Nov 16 21:02:41 kernel: et0: et_mvlan_netdev_event: event 1 for vlan1 mvlan_en 0

Nov 16 21:02:41 kernel: et0: et_mvlan_netdev_event: event 13 for vlan2 mvlan_en 0

Nov 16 21:02:41 kernel: et0: et_mvlan_netdev_event: event 1 for vlan2 mvlan_en 0

Nov 16 21:02:48 nat: apply nat rules (/tmp/nat_rules_eth0_eth0)

Nov 16 21:02:49 wan: finish adding multi routes

Nov 16 21:02:49 rc_service: udhcpc 19879:notify_rc stop_upnp

Nov 16 21:02:49 rc_service: udhcpc 19879:notify_rc start_upnp

Nov 16 21:02:49 rc_service: waitting "stop_upnp" via udhcpc ...

Nov 16 21:02:51 dhcp_client: bound 172.24.1.230/255.255.255.0 via 172.24.1.1 for 43200 seconds.

Nov 16 21:03:01 nat: apply redirect rules

Forced to reboot.
Flashed back 13 no issues.

 

[RMerlin]

About me - nothing special:
No IPV6, no DoT or DoH, only Diversion.

Try disabling Diversion, as it adds a lot of entries to dnsmasq.

The majority of us (including myself) do not experience any such issues, so there has to be something different with your setup.

 

[jsarcone]

I am using DoT and DNS Caching. I am not sure what you mean by Diversion. I have forwarders. I have no IP6. I am using NTP & the Redirect.

 

[skeal]

There is something definitely wrong. I flashed it again and after about a few hours and depending on the load; it loses connection to the Internet. I waited little longer and was able to login to Web Interface. I was able to grab logs and CPU 2 was pigged. Attempted to restart the WAN interface and it did work.

logs:

Nov 16 14:11:20 crond[643]: time disparity of 807245 minutes detected

Nov 16 20:55:06 nat: apply redirect rules

Nov 16 20:55:06 WAN_Connection: ISP's DHCP did not function properly.

Nov 16 21:02:33 rc_service: httpd 645:notify_rc restart_wan_if 0

Nov 16 21:02:34 kernel: Attempt to kill tasklet from interrupt

Nov 16 21:02:34 kernel: et0: et_mvlan_netdev_event: event 9 for vlan1 mvlan_en 0

Nov 16 21:02:34 kernel: et0: et_mvlan_netdev_event: event 2 for vlan1 mvlan_en 0

Nov 16 21:02:34 kernel: et0: et_mvlan_netdev_event: event 13 for vlan1 mvlan_en 0

Nov 16 21:02:34 kernel: et0: et_mvlan_netdev_event: event 1 for vlan1 mvlan_en 0

Nov 16 21:02:34 kernel: et0: et_mvlan_netdev_event: event 13 for vlan2 mvlan_en 0

Nov 16 21:02:34 kernel: et0: et_mvlan_netdev_event: event 1 for vlan2 mvlan_en 0

Nov 16 21:02:34 kernel: et0: et_mvlan_netdev_event: event 4 for vlan1 mvlan_en 0

Nov 16 21:02:41 rc_service: httpd 645:notify_rc restart_wan_if 0

Nov 16 21:02:41 kernel: Attempt to kill tasklet from interrupt

Nov 16 21:02:41 kernel: et0: et_mvlan_netdev_event: event 9 for vlan1 mvlan_en 0

Nov 16 21:02:41 kernel: et0: et_mvlan_netdev_event: event 2 for vlan1 mvlan_en 0

Nov 16 21:02:41 kernel: et0: et_mvlan_netdev_event: event 9 for vlan2 mvlan_en 0

Nov 16 21:02:41 kernel: et0: et_mvlan_netdev_event: event 2 for vlan2 mvlan_en 0

Nov 16 21:02:41 kernel: et0: et_mvlan_netdev_event: event 13 for vlan1 mvlan_en 0

Nov 16 21:02:41 kernel: et0: et_mvlan_netdev_event: event 1 for vlan1 mvlan_en 0

Nov 16 21:02:41 kernel: et0: et_mvlan_netdev_event: event 13 for vlan2 mvlan_en 0

Nov 16 21:02:41 kernel: et0: et_mvlan_netdev_event: event 1 for vlan2 mvlan_en 0

Nov 16 21:02:48 nat: apply nat rules (/tmp/nat_rules_eth0_eth0)

Nov 16 21:02:49 wan: finish adding multi routes

Nov 16 21:02:49 rc_service: udhcpc 19879:notify_rc stop_upnp

Nov 16 21:02:49 rc_service: udhcpc 19879:notify_rc start_upnp

Nov 16 21:02:49 rc_service: waitting "stop_upnp" via udhcpc ...

Nov 16 21:02:51 dhcp_client: bound 172.24.1.230/255.255.255.0 via 172.24.1.1 for 43200 seconds.

Nov 16 21:03:01 nat: apply redirect rules

Forced to reboot.
Flashed back 13 no issues.

I just went through my logs and nowhere do I see a reference to "vlans" like yours, I think @RMerlin is right your specific setup is likely the cause. What I mean is, a misconfiguration. Reset to defaults and add back your specific unique settings one by one until you break the system.

EDIT: Do you have Game boost turned on?

 

[PeterV]

Try disabling Diversion, as it adds a lot of entries to dnsmasq.

The majority of us (including myself) do not experience any such issues, so there has to be something different with your setup.

I made it - with no lucky.
But, as I promised, I tried to use dnsmasq from aplha2 in beta1 - and it is working like a charm.
No any issues during the whole day.
I.e. instead Dnsmasq version 2.80-93-g6ebdc95 I mount Dnsmasq version 2.80-74-gdefd6b1

 

[consorts]

changelog - OpenVPN (2.4.8)

anyone notice any performance change in router based openvpn client use?
please indicate which model router you are using if it's not in your signature.

perhaps i should clarify... i'm not interested in your vpn performance
i'm interested in any CHANGE to your performance since going 384.14
is it faster, slower, no perceptible difference, drops more, cpu% change, etc.

i'm on .13 with a ac3100 so my cpu is NOT optimized for openvpn
thus i would like to know if .14 will offer anything worth upgrading.

 

[no_name]

changelog - OpenVPN (2.4.8)

anyone notice any performance change in router based openvpn client use?
please indicate which model router you are using if it's not in your signature.

I use ExpressVPN client on my RT-AC86U, I pay for 100 mb download and get around 9 mb upload.

Sent from my iPad using Tapatalk

 

[jsarcone]

I just went through my logs and nowhere do I see a reference to "vlans" like yours, I think @RMerlin is right your specific setup is likely the cause. What I mean is, a misconfiguration. Reset to defaults and add back your specific unique settings one by one until you break the system.

EDIT: Do you have Game boost turned on?

No I don't have Game Boost turned on. Error I think is starting is Nov 16 20:55:06 WAN_Connection: ISP's DHCP did not function properly.

Then the router falls over.
I did reset the router and configure what I needed back in. However, this issue still occurred. Reflash to 13 and reapplied my save config prior to going to 14. Basically, breaking my environment to get internet access I still saw this issue. LAN > AC68 > DMZ (V-LAN'd SWITCH) > SERCUITY DEVICE> INTERNET. Note: configuration as work for years.

The error I keep getting is DHCP request from WAN port failing

 

[CaptainSTX]

changelog - OpenVPN (2.4.8)

anyone notice any performance change in router based openvpn client use?
please indicate which model router you are using if it's not in your signature.

Running both an AC1900P (384.13 ) and an AC86 384. beta) alternatively. Both set up with 2 open VPN clients and a VPN server.

On the AC1900P when doing a speedtest core 1 use jumps to 45% core 2 (VPN running on this core ) jumps to 89%. Everything continues to function for me but if I was running QOS or streaming on multiple devices there would probably be some impact, however in the eighteen months that I have used this router online I have never had an issue. I gave an AC1900P to my son so he can run a VPN client to watch out of market baseball games and he has never had an issue.

On the AC86 because of the AES-NI support there is much smaller impact on CPU usage so you have more headroom for other functions partially offset by the fact that this router can handle VPN encryption at near line speed so throughput on the VPN tunnels is 3x - 4x faster than on the AC1900P so it does load up both cores. The fast VPN is the upside of this router that makes it the go to router for people that want to run VPN clients. When I remote in using the VPN client on either the AC1900P or the AC86 really no difference between the routers as the limiting factor is that Comcast limits my upload speed to 10 Mbps.

I have also run multiple VPN clients and a VPN server on an N66. It worked and it never crashed but with people's quest for speed not really viable today for running a VPN. One important factor in performance is that Merlin has optimized VPN performance by giving you the option on which core(s) to run VPN clients.

 

[Val D.]

anyone notice any performance change in router based openvpn client use?
please indicate which model router you are using if it's not in your signature.

RT-AC86U, Asuswrt-Merlin 384.12, NordVPN local server, 300/20 ISP
I don't know how better it may get with the updated OpenVPN.

i'm interested in any CHANGE to your performance since going 384.14

I see your updated post. Sorry, no plans to update this router any time soon.
Nothing is broken on it that requires fixing. I actually rolled back to 384.12 from 384.13. Works better for me.

 

[consorts]

rolled back to 384.12 from 384.13. Works better for me.

yeah, while i love dns-tls (instead of using stubby) i'm having a few
.13 regrets myself, what exactly provoked you to roll back to .12 ?

 

[Makaveli]

RT-AC86U, Asuswrt-Merlin 384.12, NordVPN local server, 300/20 ISP
I don't know how better it may get with the updated OpenVPN.

View attachment 19924



I see your updated post. Sorry, no plans to update this router any time soon.
Nothing is broken on it that requires fixing. I actually rolled back to 384.12 from 384.13. Works better for me.

384.13 Express VPN Local server. 1Gbps/750mbps Fiber

Once 384.14 is out of beta I can provide numbers for that.

 

[no_name]

I’ve seen no change in performance, speed or cpu usage with OpenVPN (2.4.8)


Sent from my iPad using Tapatalk

 

[RMerlin]

I made it - with no lucky.
But, as I promised, I tried to use dnsmasq from aplha2 in beta1 - and it is working like a charm.
No any issues during the whole day.
I.e. instead Dnsmasq version 2.80-93-g6ebdc95 I mount Dnsmasq version 2.80-74-gdefd6b1

Paste the content of /etc/dnsmasq.conf (you can skip the dhcp-host entries which are your static leases) and /tmp/resolv.dnsmasq.

You can also try with debug logging enabled to see what queries happens just before the crash (in case it's a very specific query causing it to crash, which would also explain why it doesn't crash at all for most of us). In a /jffs/configs/dnsmasq.conf.add:

log-queries

 

[RMerlin]

changelog - OpenVPN (2.4.8)

anyone notice any performance change in router based openvpn client use?
please indicate which model router you are using if it's not in your signature.

perhaps i should clarify... i'm not interested in your vpn performance
i'm interested in any CHANGE to your performance since going 384.14
is it faster, slower, no perceptible difference, drops more, cpu% change, etc.

i'm on .13 with a ac3100 so my cpu is NOT optimized for openvpn
thus i would like to know if .14 will offer anything worth upgrading.

There were very few changes between 2.4.7 and 2.4.8, mostly bug fixes, so I wouldn't expect any significant performance difference.

 

[PeterV]

Paste the content of /etc/dnsmasq.conf (you can skip the dhcp-host entries which are your static leases) and /tmp/resolv.dnsmasq.

You can also try with debug logging enabled to see what queries happens just before the crash (in case it's a very specific query causing it to crash, which would also explain why it doesn't crash at all for most of us). In a /jffs/configs/dnsmasq.conf.add:

log-queries

Hi.
Log queries is already enabled.
Nothing special. It wasn't crashed. It use one whole core (100% of core) and wasn't responce on any DHCP queries.
In log it's look like a long time without DHCP queries.
From wifi the symptom is - it's stuck on "Receiving the IP-address".
I can't upload a dnsmasq.conf, there is a pastebin link

 

[themiron]

Hi.
Log queries is already enabled.
Nothing special. It wasn't crashed. It use one whole core (100% of core) and wasn't responce on any DHCP queries.
In log it's look like a long time without DHCP queries.
From wifi the symptom is - it's stuck on "Receiving the IP-address".
I can't upload a dnsmasq.conf, there is a pastebin link

try to remove "dhcp-script=/sbin/dhcpc_lease" from dnsmasq.conf, will it help?

syslog and /opt/var/log/dnsmasq.log will give some light on this.

 

[Sanna1967]

6 days uptime, and everything works fine. Nothing fancy, vanilla config with 1 vpn client(Xpress_vpn) + 1 mesh node.

*except for some debugging log messages here and there.I think we will have to live with

 

[RMerlin]

try to remove "dhcp-script=/sbin/dhcpc_lease" from dnsmasq.conf, will it help?

Also if you get it to crash again, run "ps w" over SSH and post the list of running processes, so we can see if something else is stuck at high CPU usage.

I experimented a bit with a bogus dhcpc_lease command here, so far I still can't reproduce your issue either.

 

[CriticJay]

any chance that's Esso, and the Extra app rather than SpeedPass (which I don't use)?

Correct, fellow Poutine-eater.