Release Asuswrt-Merlin 388.2 is now available for select models

[plexion]

All cores do get used, you can see them with top. Core usage will vary by affinity, some processes like OpenVPN are locked to specific cores.

If something has high CPU usage then you need to track it down using top.

Thanks for the pointer @RMerlin.

I see only 1 process with CPU3 on 388.2 in the top output, but more processes assigned to CPU3 on 386.7_2.
What might be the reason of this change?

As for the higher CPU usage, I noticed the process - sysstate is using up a lot of CPU cycles.
I have a USB HDD attached for FTP use, and the high CPU usage goes away if I unmount the disk.

Any idea what might be the cause please?

Thank you.

 

[Piotrek]

As for the higher CPU usage, I noticed the process - sysstate is using up a lot of CPU cycles.
I have a USB HDD attached for FTP use, and the high CPU usage goes away if I unmount the disk.

Any idea what might be the cause please?

This is a router with a low CPU, not a server with many powerful CPUs.
Use the router function, not the server function.

 

[plexion]

This is a router with a low CPU, not a server with many powerful CPUs.
Use the router function, not the server function.

I understand that it is a router.
The thing is it is working perfectly fine up to 386.7_2 if you see my previous posts, and the FTP server is not getting high activity.
That is why I am trying to understand what is causing the issues on 388.2.

 

[Piotrek]

That is why I am trying to understand what is causing the issues on 388.2.

Maybe your USB flash drive is running out of life?
The 388 firmware needs more resources than the 386, maybe such a simple router is not powerful enough to handle non-essential functions?

 

[jmat]

Thanks for the pointer @RMerlin.

I see only 1 process with CPU3 on 388.2 in the top output, but more processes assigned to CPU3 on 386.7_2.
What might be the reason of this change?

As for the higher CPU usage, I noticed the process - sysstate is using up a lot of CPU cycles.
I have a USB HDD attached for FTP use, and the high CPU usage goes away if I unmount the disk.

Any idea what might be the cause please?

Thank you.

It is possible that the USB disk is causing you a problem. In my case, I mounted a USB in an AX86U only for swap, and it caught my attention that the ram level had risen to almost 100% of a total of 1024 MB, and on the other hand I began to have cuts in the Wi-Fi signal, although I did not observe any changes or any anomalies in the use of the CPU cores. I have now unmounted the USB and everything seems to be back to normal, with ram usage around 50%...

 

[saison2023]

@saison2023 had the same results I did. Regarding Diversion lite/pixelserve… Didn’t I already reported that back to you? Because that’s what I have already confirmed: it’s NOT pixelserve. It’s the logging of Diversion (as that’s switched on by default) in combination with dnsmasq 2.89

This isn't dnsmasq crash. It's related to httpds probably caused by pixelserve. Try diversion lite and see what happens.

I already tried to disable pixelserv and to use Diversion lite without any success.
As @Bogey mentionned, we already tried a lot of things and it always come back to dnsmasq logging.

 

[jmat]

It is possible that the USB disk is causing you a problem. In my case, I mounted a USB in an AX86U only for swap, and it caught my attention that the ram level had risen to almost 100% of a total of 1024 MB, and on the other hand I began to have cuts in the Wi-Fi signal, although I did not observe any changes or any anomalies in the use of the CPU cores. I have now unmounted the USB and everything seems to be back to normal, with ram usage around 50%...

PS: Although at least in my case I think the Wi-Fi outages are more related to having the <Smart Connect> option enabled, a problem that did not occur to me in versions prior to 388.2

 

[RMerlin]

I see only 1 process with CPU3 on 388.2 in the top output, but more processes assigned to CPU3 on 386.7_2.
What might be the reason of this change?

I'm not aware of any changes at this level. CPU assignation might simply depend on the actual CPU load - no point in assigning to core 3 if core 0 through 2 are already freely available when the process gets scheduled to run.

That router has always had minimal CPU usage on core 3, you can see threads discussing that as far as a few years ago.

 

[bluepoint]

I already tried to disable pixelserv and to use Diversion lite without any success.
As @Bogey mentionned, we already tried a lot of things and it always come back to dnsmasq logging.

Your crash has to do with diversion logging related to httpds which I don't experience. If that's the case and you seems to be satisfied with your findings I'll be staying in the sideline. Good luck.

 

[plexion]

I'm not aware of any changes at this level. CPU assignation might simply depend on the actual CPU load - no point in assigning to core 3 if core 0 through 2 are already freely available when the process gets scheduled to run.

That router has always had minimal CPU usage on core 3, you can see threads discussing that as far as a few years ago.

Thanks for the explanation @RMerlin.

How about for the high CPU usage by the "sysstate" process when a USB HDD is mounted?
Any idea on that?

Thanks.

 

[RMerlin]

How about for the high CPU usage by the "sysstate" process when a USB HDD is mounted?
Any idea on that?

No idea. Closed source component.

 

[plexion]

No idea. Closed source component.

Alright, thank you.
Guess I have to stay on 386.7_2 for the moment as I need the USB HDD to be connected.

 

[saison2023]

Your crash has to do with diversion logging related to httpds which I don't experience. If that's the case and you seems to be satisfied with your findings I'll be staying in the sideline. Good luck.

Thanks but I did not want to sound rude, just to explain that we did a lot of work before sharing.
But thanks again for your help.

 

[vlord]

So is that what’s causing the device/computer disconnects I’m experiencing? (And if I’m lucky to have still one device connected I could restart dnsmasq so they can reconnect.)

I think I was experiencing this as well. You can read about it in this thread. I thought it might be my chatty roku and amazon devices + having the large oisd.nl list in place but very eerily similar where devices would go offline. Reconnects, if they’d even work, would take 5-10 mins. Often, I’d just reboot the router to see if I could get it fixed. Ultimately switched over to AGH and issues went away.

 

[SomeWhereOverTheRainBow]

Every since switching to the lastest version (388.2_2) my line speed has gone through the roof!

So this is from a client...

None of my network clients have 2.5Gbps lan cards, I currently have flexQoS installed as well.

 

[RMerlin]

Every since switching to the lastest version (388.2_2) my line speed has gone through the roof!

View attachment 50209

That difference is typically indicative of going from 1 Gbps to 2.5 Gbps.

 

[SomeWhereOverTheRainBow]

That difference is typically indicative of going from 1 Gbps to 2.5 Gbps.

I guess my ISP heard I updated the router firmware and decided to give me the Merlin package without telling me. I am going to look closely at my bill the next couple of months hahaha. Those are some nice numbers. It will be even better once Asus starts doing everything 2.5Gbps. BTW those numbers are from the RT-AX88U Pro.

 

[SomeWhereOverTheRainBow]

Ultimately switched over to AGH and issues went away.

You and @Stephen Harrington are going to start a trend......

 

[Damit1]

10 days a row, rock f****g solid.

 

[hancox]

I guess my ISP heard I updated the router firmware and decided to give me the Merlin package without telling me. I am going to look closely at my bill the next couple of months hahaha. Those are some nice numbers. It will be even better once Asus starts doing everything 2.5Gbps. BTW those numbers are from the RT-AX88U Pro.

"Working" QoS (with fc on), and getting those speeds, is noteworthy. The AX86U (not pro) can't do it, and Asus doesn't seem to care.