XR 500 CPU Spikes and High CPU

[Chrisvw]

Both have open NAT. Game lagging as in player  gets stuck  while running in game, then if your in party chat you get disconnected from the party. Message on console top right hand corner  some network features no longer available. Can no longer join any party chat. Reboot PS4 unable to sign in PS network. Once you reboot router all is back to normal.

[Netduma Jack]

6 hours ago, Chrisvw said:

Both have open NAT. Game lagging as in player  gets stuck  while running in game, then if your in party chat you get disconnected from the party. Message on console top right hand corner  some network features no longer available. Can no longer join any party chat. Reboot PS4 unable to sign in PS network. Once you reboot router all is back to normal.

That sounds really weird. Does this happen with the Geo-Filter disabled? i.e no devices added to it? Try to rule out which feature might be causing this to happen. I don't think it'll be caused by having two PS4's as such, rather there's probably something configured incorrectly. Do you use UPnP as well, or are you manually port forwarding?

[Chrisvw]

Both have open NAT. Game lagging as in player  gets stuck  while running in game, then if your in party chat you get disconnected from the party. Message on console top right hand corner  some network features no longer available. Can no longer join any party chat. Reboot PS4 unable to sign in PS network. Once you reboot router all is back to normal.

[Chrisvw]

12 hours ago, Netduma Jack said:

That sounds really weird. Does this happen with the Geo-Filter disabled? i.e no devices added to it? Try to rule out which feature might be causing this to happen. I don't think it'll be caused by having two PS4's as such, rather there's probably something configured incorrectly. Do you use UPnP as well, or are you manually port forwarding?

No features enabled.

[Netduma Jack]

6 hours ago, Chrisvw said:

No features enabled.

Have you ever manually changed your DNS? Change that to 8.8.8.8 and see if that sorts it. If not, try a factory reset in case something's gone wrong with the settings. I'm leaning towards this being a PS4 bug rather than anything to do with your router.

[Chrisvw]

Hi I have done a factory reset. Yes I have tried a DNS change, this is a router issue when I use my TP Link AC 1750 I don't get this issue.

[Guest Killhippie]

57 minutes ago, Chrisvw said:

Hi I have done a factory reset. Yes I have tried a DNS change, this is a router issue when I use my TP Link AC 1750 I don't get this issue.

This does not help but I don't get this issue with the XR500 and we have not seen a flood of this issues either so its something unique to your setup. I would start with a serious virus scan, download something like Malwarebytes as well so you have an active and passive scanner if its a windows machine. Also is the behaviour on the latest firmware 2.3.2.40? You can only get it by going into settings > administration and pressing > firmware update  and then the check for new firmware.

[Chrisvw]

This is only PS4 related streaming and web browsing is not affected.

[Guest Killhippie]

4 minutes ago, Chrisvw said:

This is only PS4 related streaming and web browsing is not affected.

Have you updated to the latest XR500 firmware that came out this week? It has a stability fix from Qualcomm and you wont be vulnerable to the security issue 2.3.2.22 has which isn't pretty. Also a virus scan cant hurt, its just ticking the t's and dotting the i's so to speak.               

[Chrisvw]

1 minute ago, Killhippie said:

Have you updated to the latest firmware that came out this week? It has a stability fix from Qualcomm and you wont be vulnerable to the security issue 2.3.2.22 has which isn't pretty. Also a virus scan cant hurt, its just ticking the t's and dotting the i's so to speak.               

No PC running when we game and stream but will do in any case. Updated the new router when it was delivered to the latest version.  I really like the router when it works just want to get this sorted.

[Guest Killhippie]

4 minutes ago, Chrisvw said:

No PC running when we game and stream but will do in any case. Updated the new router when it was delivered to the latest version.  I really like the router when it works just want to get this sorted.

I can understand your frustration, that's why I suggest you update your firmware, there is a new version that came out this week. You are running 2.3.2.22, yes? The latest version is 2.3.2.40. Please update as this may help your issue, you should always when possible run the latest firmware.

[Guest Killhippie]

I hope it helps, fingers crossed. Lots of bug fixes and new features too.

[Chrisvw]

Updated lets see how it goes

 

[Guest Killhippie]

Thats about how mine is, the version new one uses a tad more CPU tthan the last version.

[GazCBG]

Mine is between 40 and 60 for both CPU's however I am hammering my connection, as I have had to download all my PS4 games.
Since the firmware update I have done 2.2TB on download on a 220Mbps download connection non-stop (09:05:56 up 1 day, 10:43, load average: 4.61, 4.62, 4.60).

 

I used this site: https://convertlive.com/u/convert/bytes/to/gigabytes to convert the Bytes from Dashboard (Network Status)

[Guest Killhippie]

9 minutes ago, GazCBG said:

Mine is between 40 and 60 for both CPU's however I am hammering my connection, as I have had to download all my PS4 games.
Since the firmware update I have done 2.2TB on download on a 220Mbps download connection non-stop (09:05:56 up 1 day, 10:43, load average: 4.61, 4.62, 4.60).

 

I used this site: https://convertlive.com/u/convert/bytes/to/gigabytes to convert the Bytes from Dashboard (Network Status)

On my unit core 1 may be running at say 10% then spike to 40-60% for no reason then stay at 40% briefly while Core 2 will drop to about 10-20% then core 2 will spike to 40-60% and drop back to 40% and core 1 drops back down to around 10%, and the cycle continues with both cores swapping over that way. I tried speedtests and downloads and tbh the graph does not change much, which is a tad odd. It only spikes to almost 100% when opening a new GUI menu, or going back to the dashboard.

 I find loading the GUI maxing out both cores bizarre when its so hard to tax them that hard normally, especially when the network cores (the two 800Mhz cores) are also doing their thing to accelerate packet processing, unless netgears firmware is not utilizing all four cores properly. The R7800 was underlocked at 1.4Ghz, Voxel had to adjust the code to make the Krait cores run at 1.7Ghz. Netgear has a habit of either using one core of a dual core CPU, not using all the ram or underclocking  both cores. The IPQ8065 is a big.LITTLE Soc, it would be nice to see what the network cores are doing too, as  well as the two 1.7Ghz Krait cores. https://www.qualcomm.com/products/ipq8065

[GazCBG]

That is what mine was like on the old firmware, but due to the downloading of the PS4 games (should be finished today), it not had time to settle.
I also noticed the GUI also spiking the CPU's.
What I had noticed is my load average is always around 4 when I logged in to the router, not sure if that normal or not.

[xr500user]

10 hours ago, Killhippie said:

On my unit core 1 may be running at say 10% then spike to 40-60% for no reason then stay at 40% briefly while Core 2 will drop to about 10-20% then core 2 will spike to 40-60% and drop back to 40% and core 1 drops back down to around 10%, and the cycle continues with both cores swapping over that way. I tried speedtests and downloads and tbh the graph does not change much, which is a tad odd. It only spikes to almost 100% when opening a new GUI menu, or going back to the dashboard.

 I find loading the GUI maxing out both cores bizarre when its so hard to tax them that hard normally, especially when the network cores (the two 800Mhz cores) are also doing their thing to accelerate packet processing, unless netgears firmware is not utilizing all four cores properly. The R7800 was underlocked at 1.4Ghz, Voxel had to adjust the code to make the Krait cores run at 1.7Ghz. Netgear has a habit of either using one core of a dual core CPU, not using all the ram or underclocking  both cores. The IPQ8065 is a big.LITTLE Soc, it would be nice to see what the network cores are doing too, as  well as the two 1.7Ghz Krait cores. https://www.qualcomm.com/products/ipq8065

i looked at the cpu krait its set to 'boost', but i have no idea if this is what it should be

all cpufreq setting files i found were set 1750000 as clockrate max, but they do have min and low values set, i saw one that had 1475000 which may be the underclocking your talking about , but i have no way of telling if its in effect unless i install things on the router, too new for me to mess with it yet - the kernel is new though (nov 29th build) but no idea how it was built and what options were checked regarding cpu, there may be logs but i haven't looked. it's an android kernel and chipset, so that's why they have those other 2 800 mhz cores on qualcomms chip.. which if its being run on a phone you know it uses the slower cores to save battery power, or show an always on display etc. not sure if they are being used and even if they would add any benefit if they were used? lots of things are hardware based on the router board, like hardware chips to accelerate packet inspection, hardware switch for port 1,2,3,4, wan & lan, there r lots of mini processors in there doing different things Independently from the main cpu cores

the scaling is set to 'ondemand' type https://wiki.archlinux.org/index.php/CPU_frequency_scaling#Tuning_the_ondemand_governor

tweaking here can bring it up, but has to do with power savings.. but see below anyway ..

I went way deeper into the cpu usage app.  it's sorta bs man - don't listen to it, its showing load and not cpu usage based on idle and wait state with a mathematical function to determine it and even display it.. and i think its not really accurate, the jump is because your browser is influencing the graphic, i bet different browsers on diff machines will show diff things..  even when you click the ? in the rapp it says reporting amount of processes "The number(s) will be higher when there are more processes running on your router. " -- which is load, not necessarily the cpu usage (two different things) and processes in wait or sleeping , suspended, or zombies also count towards these #s .. they have added more processes since the last update, so the number is higher influencing the code.

uptime load #s are  1 minute,5minutes, 15minutes -- and unreliable .. but again its load - not cpu usage. the graphic is not reflecting actual cpu usage as far as i can tell and i think i know why - its a javascript file, and i tried to make some changes to make it show exactly the cpu load but i think it would be a very boring and disappointing graphic then, its eye candy imo.. i wound up getting the window to scale so the lower the cpu was the closer the zoom, and it was close to actual cpu usage, then i set everything back because i realize its just eye candy

i watched the cpu's, even with 40 devices , one running speed tests, and constantly refreshing duma os dashboard i was lucky to see cpu usage jump past 3% lol

once it went to 24% then back down to 2%.. most processes are sleeping, or using 0.01%cpu maybe 20 of them - it would make for a very boring r-app .. unless i'm missing something, i would not rely on that for cpu usage -- load, yes .. but its high because of a lot of processes being created/destroyed or fork()'s that's all.

 

 

 

[Guest Killhippie]

I guessed it was the browser, I tried with safari mobile and it looked different, also if you dont lear your cache the Dashboard loads faster, Netgeasr never really relied on so much on the browser cache in previous routers  so much, I came to the same conclusion as you, its there to look pretty really, a touch of eye candy but load is what counts. Interesting to know the build is that new, I was informed it was delayed by Netgear before final release so this has to be one of Netgears most up to date releases, which is a good thing tbh. Maybe they will start updating packages soon rather than botching old ones!

 I think the 800Mhz cores on the CPU are just doing packet acceleration in the background, nothing special, I know something about them doing tasks like offload processors rather than separate offload CPU's actually on the PCB. I think cpufreq 1750000 is its maximum clock rate and maybe Netegar have a resting state now ie; 1475000. The R7800 was clocked at 1.4Ghz and Voxel changed that to get more out of the router,  its also possible the cpufreq 1475000 is just old code from the R7800 days, Netgear do leave dead lines of code in the firmware at times. Its like re writing a book but not removing the older story lines.

 The XR500 is running well for me I have to say on the latest firmware, which is still not available via Netgears firmware page, it would be great to see the release notes as it does have a Qualcomm hotfix. Also wondering about any security updates. Did you find what version of OpenSSL its running by any chance? I cant see that what the current version is. With no published GPL code (again) for this update it makes getting to the bottom of all the changes a real pain at times.

[xr500user]

yeah. .32 kernel was built mid August, this one is late November.

there is some validity to the graphic, but i'm sure it's measuring load (number of processes), and load can be huge if most of it is sleeping and not doing anything really with the cpus. server load and cpu load/usage two diff things - but  i watched every second refreshing and it was hardly using any cpu - 2-5% any time i looked -- i assume it jumps when a lot of stuff is going on, but i haven't seen it go past  24% for a split second. it has plenty of power, it can probably do the plex server too and much more with no impact

the versions are reporting

OpenSSL 1.0.2n  7 Dec 2017
built on: reproducible build, date unspecified
platform: linux-armv4-openwrt
options:  bn(64,32) rc4(ptr,char) des(idx,cisc,2,long) blowfish(ptr)
compiler: arm-openwrt-linux-uclibcgnueabi-gcc -I. -I.. -I../include  -fPIC -DOPENSSL_PIC -DOPENSSL_THREADS -D_REENTRANT -DDSO_DLFCN -DHAVE_DLFCN_H -I/home/pengyu.zhou/works/git_repos/gaming_router-buildroot-256/staging_dir/target-arm_v7-a_uClibc-0.9.33.2_eabi/usr/include -I/home/pengyu.zhou/works/git_repos/gaming_router-buildroot-256/staging_dir/target-arm_v7-a_uClibc-0.9.33.2_eabi/include -I/home/pengyu.zhou/works/git_repos/gaming_router-buildroot-256/staging_dir/toolchain-arm_v7-a_gcc-4.6-linaro_uClibc-0.9.33.2_eabi/usr/include -I/home/pengyu.zhou/works/git_repos/gaming_router-buildroot-256/staging_dir/toolchain-arm_v7-a_gcc-4.6-linaro_uClibc-0.9.33.2_eabi/include -DOPENSSL_SMALL_FOOTPRINT -DOPENSSL_NO_ERR -DTERMIOS -Os -pipe -march=armv7-a -mfpu=vfpv3-d16 -mfloat-abi=softfp -fhonour-copts -Wno-error=unused-but-set-variable -msoft-float -fpic -I/home/pengyu.zhou/works/git_repos/gaming_router-buildroot-256/package/openssl/include -ffunction-sections -fdata-sections -fomit-frame-pointer -Wall -DOPENSSL_BN_ASM_MONT -DOPENSSL_BN_ASM_GF2m -DSHA1_ASM -DSHA256_ASM -DSHA512_ASM -DAES_ASM -DBSAES_ASM -DGHASH_ASM

OpenVPN 2.4.3 arm-openwrt-linux-gnu [SSL (OpenSSL)] [LZO] [LZ4] [EPOLL] [MH/PKTINFO] [AEAD] built on Nov 29 2018
library versions: OpenSSL 1.0.2n  7 Dec 2017, LZO 2.06
Originally developed by James Yonan
Copyright (C) 2002-2017 OpenVPN Technologies, Inc. <[email protected]>

looks like openssl is a year old, but no idea if it was modified from what it is reporting, i am guessing openvpn is what hybrid vpn is using

[Chrisvw]

Ok it ran for 2 days and it happened again. Is there any way I can get any other log files see where this is happening? Busy playing but not lagging anymore just message network features not available then disconnected from the servers have to reboot to be able to connect.

[Guest Killhippie]

On 12/16/2018 at 5:33 PM, xr500user said:

yeah. .32 kernel was built mid August, this one is late November.

It looks like openssl is a year old, but no idea if it was modified from what it is reporting, i am guessing openvpn is what hybrid vpn is using

I cant believe they have not updated OpenSSL since 1.0.2n. This branch should be now 1.0.2q to fix a low level CVE. Voxel does when new versions are available for the R7800. This is something netegar ignore, I don't think they can patch it like miniupnd, it just needs the latest included with new they updates The latest version is OpenSSL 1.1.1a , including bug and security fixes. Here's the current list of CVE's since a year ago. on the 1.0.2 branch https://www.openssl.org/news/vulnerabilities-1.0.2.html
Netgear really dont try hard to keep the routers secure. I wish Netduma's team could keep this up to date smh.