100% cpu usage on core 1 of AC88U

[curlysir]

The only thing that fixed it for me was a complete power off. A reboot will not do. I had to do it twice, once for each core. Have not had the problem since.

 

[Stpijb]

The only thing that fixed it for me was a complete power off. A reboot will not do. I had to do it twice, once for each core. Have not had the problem since.

I tried that 4 times. Unplugged the router and let sit for 1min at a time. Eventually I got tired of it and changed to the legacy firmware and its running like a champ.

 

[TheLyppardMan]

Hey guys, thanks for the awesome work you put into this firmware. I just purchased an ac88u and installed the latest and greatest stable version. I did find that core 1 was using 70-100% while core 2 sat almost completely idle. I couldn't even login with putty to see what was going on. I did try a lot of things and rebooted after each method to fix. First i turned wps off didn't fix it. Turned qos off didn't fix. Turned smart connect off didn't fix. Turned guest network off didn't fix. Turned 5ghz off didn't fix. Turned 2.4ghz off didn't fix.

After all of this I downloaded the legacy firmware reset the router and installed the latest legacy version. Usage dropped to 1% across both cores. I didn't see memory usage ever go up. So something is acting up with the stable versions.

Obvious question, but did you perform a factory reset immediately after flashing with Merlin firmware and before making any configuration changes?

 

[Stpijb]

Obvious question, but did you perform a factory reset immediately after flashing with Merlin firmware and before making any configuration changes?

yes. I pulled a brand new router out flashed it immediately to Merlins. held the pretty little red button in to reset. I tried resetting 2 more times. and didn't solve the issue. So i re-flashed to legacy and reset the router and bam worked like a charm. I spent about 4 hours playing with it yesterday before throwing the towel in and going to legacy version.

I dont think its merlins firmware I think it something Asus has bugged. But since Merlin rides the Asus firmware its also having the problem effect it.

 

[RMerlin]

Tried that also. Set 2 and 157. Made it worst with core 1 stock at 100.

ACSD is the Automatic Channel Selection Daemon from Broadcom. If channels are all manually set, then this shouldn't be running.

admin@Stargate88:/tmp/home/root# ps w | grep acsd
23912 admin     1428 S    grep acsd

 

[Stpijb]

ACSD is the Automatic Channel Selection Daemon from Broadcom. If channels are all manually set, then this shouldn't be running.

admin@Stargate88:/tmp/home/root# ps w | grep acsd
23912 admin     1428 S    grep acsd

Yes I turned smart connect and the auto channel select off. Had to manually set the channels for both bands.

 

[BSOD2600]

[QUOTE="RMerlin][/QUOTE]

@RMerlin
What additional info, debug logs, etc do you need to track down the 50% IO usage due to USB drives, several of us are experiencing

 

[sfx2000]

I've been using the Merlin firmware for my router for quite a while now, and just today I noticed that core 1 on my Asus AC88U is pegged at 100% all of the time, and restarting the router does not solve the issue.

I wouldn't worry too much about this - as top can be misleading.... 100 percent CPU utilization means that the core is working, and not wasting time/resources...

it's not "too much CPU usage"

Putting in in lay terms - top is bad to determine things - could be that the CPU is stalled waiting on data or instructions, but that thread is committed until things move forward.

Some of this is Broadcom's code, and some of it is Asus and Third Party stuff - that being said, there has to be a certain amount of trust that Broadcom and Asus, both have done due diligence to ensure that performance meets requirements.

Asus would do users a favor by making that screen go away...

 

[TikiG]

I wouldn't worry too much about this - as top can be misleading.... 100 percent CPU utilization means that the core is working, and not wasting time/resources...

it's not "too much CPU usage"

Putting in in lay terms - top is bad to determine things - could be that the CPU is stalled waiting on data or instructions, but that thread is committed until things move forward.

This is not an accurate statement as when I experience 100% cpu usage on core 2, I cannot login to my router except via https:8443 and have other issues.

@RMerlin

I am currently in that state with core 2 at 100%. Anything you want me to try? I cannot SSH as "Network Error: Connection Refused". It is configured to be accessible on the lan side. (Note: Only after a hard reboot, I can access SSH).

RT-AC66U_B1, Running 384.4_2, QOS Script Enabled

 

[kfp]

This is not an accurate statement as when I experience 100% cpu usage on core 2, I cannot login to my router except via https:8443 and have other issues.

@RMerlin

I am currently in that state with core 2 at 100%. Anything you want me to try? I cannot SSH as "Network Error: Connection Refused". It is configured to be accessible on the lan side. (Note: Only after a hard reboot, I can access SSH).

RT-AC66U_B1, Running 384.4_2, QOS Script Enabled

Since you can login through web Ui, anything stands out in the logs?

 

[TikiG]

Issue was noticed yesterday 6/18, so logs look pretty clear. I did run a bandwidth speedtest at some point earlier in the day. Not sure about the rc_service today 6/19?

Jun 14 10:00:28 disk_monitor: Got SIGALRM...

Jun 14 10:11:13 kernel: htb: htb qdisc 12: is non-work-conserving?

Jun 14 13:07:09 kernel: htb: htb qdisc 12: is non-work-conserving?

Jun 14 23:00:29 disk_monitor: Got SIGALRM...

Jun 15 03:00:01 adaptive QOS: Scheduled Persistence Check -> No modifications necessary

Jun 15 06:59:25 kernel: htb: htb qdisc 15: is non-work-conserving?

Jun 15 09:09:01 kernel: htb: htb qdisc 13: is non-work-conserving?

Jun 15 17:58:47 kernel: htb: htb qdisc 13: is non-work-conserving?

Jun 16 03:00:01 adaptive QOS: Scheduled Persistence Check -> No modifications necessary

Jun 17 03:00:00 adaptive QOS: Scheduled Persistence Check -> No modifications necessary

Jun 18 03:00:00 adaptive QOS: Scheduled Persistence Check -> No modifications necessary

Jun 19 03:00:01 adaptive QOS: Scheduled Persistence Check -> No modifications necessary

Jun 19 11:54:01 rc_service: httpds 285:notify_rc restart_usb_idle;restart_time;restart_upnp;

Jun 19 11:54:01 kernel: klogd: exiting

 

[kfp]

Issue was noticed yesterday 6/18, so logs look pretty clear.

Random thought, could you enable telnet and login that way instead of SSH?

 

[TikiG]

Random thought, could you enable telnet and login that way instead of SSH?

Not to sound like an id10T but where is the telnet option. I looked around but did not find it under any of the options.

 

[TikiG]

Also my temperature for cpu is 154F or 67C is this normal. Just wondering if this is hot or above avg . Ambient room temp is 74F.

Thanks.

 

[kfp]

Not to sound like an id10T but where is the telnet option. I looked around but did not find it under any of the options.

Never mind, Merlin disabled telnet for security reasons. I was just thinking that a telnet connection might still work because of a lower memory requirement than SSH (assuming that’s why SSH is failing).

Temp looks normal.

 

[sfx2000]

This is not an accurate statement as when I experience 100% cpu usage on core 2, I cannot login to my router except via https:8443 and have other issues.

That doesn't wash, sorry... in a dual core device, there's ton's of threads, and you've got two core... maybe your dropbear process is hung somehow, but one must ask, why do you need to login to the dropbear in the first place?

@kfp - telnet is disabled as it's extremely insecure... it's not a memory thing, if anything, dropbear would use incrementally more...

@TikiG - what else do you have running? Services exposed to the WAN.... running entware or other package managers?

 

[kfp]

it's not a memory thing, if anything, dropbear would use incrementally more...

I never said it’s disabled because of memory footprint, and I pointed out it’s disabled because of security reasons right before your comment..

With regards to memory footprint, that’s exactly my point. Since dropbear would require more memory than telnet, telnet might have worked if it was still available and if it was really a matter of not enough memory to fork.

 

[TikiG]

That doesn't wash, sorry... in a dual core device, there's ton's of threads, and you've got two core... maybe your dropbear process is hung somehow, but one must ask, why do you need to login to the dropbear in the first place?

@kfp - telnet is disabled as it's extremely insecure... it's not a memory thing, if anything, dropbear would use incrementally more...

@TikiG - what else do you have running? Services exposed to the WAN.... running entware or other package managers?

Nothing else exposed to WAN. No entware or other package managers. FreshJr Qos Script enabled.

Dropbear should be working. I should be able to log into SSH as it is a basic function, cli and diagnosis. **Edit: Additional comment removed as it was not helpful to discussion.

 

[bsdsource]

maybe your dropbear process is hung somehow, but one must ask, why do you need to login to the dropbear in the first place?

One doesn't need to ask because it's none of your business. Dropbear should be working and shouldn't hang.

Everyone knows telnet isn't secure. It was previously disabled by default. I personally used it on occasion to run a few commands from the command line with internet disabled. Does come in handy in some situations.

This 100% CPU issue is an actual issue. If you read this entire thread you can see the issues users are experiencing.

 

[philippeannet]

This 100% CPU issue is an actual issue. If you read this entire thread you can see the issues users are experiencing.

Yep... same here... latest Merlin WRT, same issue... only help is cold (re)start
On my RT-AC88U, it's always CPU1 that's at 100%... (CPU temp 78 or 79°)