can you tell me why docker is slow and why cpu3 is 100%

[comet424]

so i mentioned before in the past i notice if unraid sits for a few hours or more  with drives powered down.. Docker is very slow.. sometimes takes a minute or 2 to populate.. just sits on  the unraid  processing  icon 

and i was using a nvme

 

so now i was able to get a 2nd ssd..  i put the nvme  all the VMs  and i put the system,docker, appdata,  on the sata ssd.. in hopes it solved that slow or sometimes doesnt load the docker..    well i was wrong still hangs  trying to load docker..

 

also i notice at times  in the pic  the one CPU is pinned at 100%  can you tell me what one is doing it..  and why not the HT also..

 

and maybe someone can tell me how to properly set the cpu isolation..  did i do it right...    should plex also get more then 1 core.. as if i do   fix common issues doesnt like that..

 

maybe i can get input what i should fix to get most optimazation.. and whats causing the docker to not really load.. i figured running 2 ssds would solve it

 

 

tower-diagnostics-20220108-1130.zip

[comet424]

right now its been sitting at this screen for 20 min...and a fresh of the screen doesnt help  just doesnt populate. and i find this after i have my drives powered down   etc

but when you normally do a reboot  then the docker is fast again pretty much...

 

and what happens when i do finally get it  icons  of dockers are gone and are replaced with   a circle ?  inside the circle

Edited January 8, 2022 by comet424

[Squid]

If the Unraid spinner doesn't disappear after 120 seconds (or 480seconds on 6.10 when it goes stable), then it will never disappear

 

The icons showing "?" are what the actual problem is.  The docker system if the icons do not exist locally attempts to download them.  And something in your network is blocking that.  Are you using piHole or something?  Try setting static DNS addresses of 208.67.222.222 / 208.67.220.220 instead of using your router as the DNS server.

 

 

[comet424]

ah  ok.. it did eventually..

i had 10 ? icons with the circle..  and if you give it time  and refresh the icons all go normal. again.. i only had 6 dockers runnin

2 lancaches,,  deluge,plex,bin krusader and transmission that was paused..

 

the icons  were none of the ones that were running..... so  the icons should be locally?  the ones in the diagnostic..  those are all installed  to the ssd.. and i been having over a year this issue were docker seems to get stuck  i get ?  but then it fixes itself...  or after a clean reboot.. docker is fast.. or if i i use unraid a bunch seems fast.. but this issue i get if i let it idle say over night  or a while and where the drives spun down  and i know i made a few posts over a year or more  saying if the docker goes to sleep    as its like it takes time to spin up even on a ssd

 

no i dont use pihole..  i use   Pfsense router with  NordVPN   and unraid goes directly through  pfsense vpn...     but anything i download goes through the 2 lancaches

they both the same.. just 1 is for VPN side  and other is for non vpn clients

 

so my clients on my network at home 

point to 192.168.32 for nonvpn   and 192.168.0.33 for vpn clients..    using lancache app from the community apps

and the upstream for both is 192.168.0.1

 

nordvpn  dns uses the dns of nordvpn

 

can you tell from my configuration diagnostics  why the cpu  is pinned on that one?   can u see my configuration settings of the dockers?

 

also  i know i followed a video when i first got unraid like 2 yrs ago  something about turning off updates to docker and plugins  so its supposed to speed it up. so docker shouldnet use internet.. just the updater plugin   the check for plugins and dockers at 12am every night and then updates..

 

 

but at moment the docker is working fast again io didnt reboot after it came back on  there was like 6 ? circles ones  and i think within 30 min as i wasnt near the computer the icons returned to normal

 

ive seen this in the past..  i have a slow connection of 5mbps   and if transmission is downloading its happened..    transmission  goes  ? circle.. but then unraid fixes it after a while..   as i download directly to a regular spinning and made it a cache pool so array stays offline..

 

hopefully that helps..    and maybe is docker checking for  updates even though should be off?

 

[Squid]

Yeah, once the icons are there they aren't downloaded again unless

1. You recreate the docker.img (all of them need to download again)
2. You change the icon URL or the name of the template (only that icon will need to re-download)
3. You add another app (only it will need to download when going to the docker / dashboard tab)

To minimize problems with the tab, there is a timeout of 45 seconds (and a 15 second connection) to grab them.  If it can't do it in that time then it aborts and next time you hit the docker tab it starts the process over again.

 

If you have in Settings - Notification Settings the check for docker updates disabled, then the OS will not check for updates. 

 

However, whenever you go to the CA tab it will automatically start a background process to check all your installed apps (plugins and docker) for updates if it hasn't been already checked by CA within the last hour.  This can be disabled in CA's settings.

 

Additionally (but not directly related), many plugins will also automatically check for updates available to them (in the background) whenever you hit their page or settings.  This includes every single one of mine, UD, parity check tuning and one Dynamix branded) - IE: Whenever you go to Apps, it will always check for updates to itself.  This behaviour cannot be disabled

[comet424]

ah ok so my issue doesnt fall into any of the above comments then.. cuz i been dealing with this issue more then a year...

so 1 2 3   do not apply..  CA didnt apply as i didnt goto it.. nore the plugins..... and how comen some of the plugins check even when u set it not check up

 

so to recreate what i go through..

1.. is i close browser.

2..wait over night say

3.. load up the browser

4.  goto    /tower

5, choose Docker

and at this point it takes the 20 min  lot of times less..  sometimes i get the ?    sometimes i dont.. sometimes its on randomly docker apps

sometimes its always transmission..   but it takes along time for tue unraid spinner

 

but if i using unraid flipping back  docker and main or share.. or using the array... the Docker page can be always fast even when my internet is bogged down

and the going to CA or pluggins doesnt apply to what i normally go through as i dont usually goto CA  or plugins  just docker to turn on and off transmission or krusader.. and then i asked few times..

 

could  Plex cause it with this giant database 

 

i wanted to know  could  my issues be  that Plex  is in the appdata...   i was thinking and i guess you cant duplicate my issue

 

but would it be better

 

ssd1   =  appdata,system,docker

ssd2 = vms

ssd3 = appdata of plex

 

as i was always thinking  plex is cause cuz it seems to be always slow loading up even on the nvme...  

 

and i have never found a solution

 

 

[comet424]

oh and i do notice i mentioned in the past...

 

plugins page will populate  faster then  docker  when docker is an issue

 

usually docker will populate like in 1 -3 seconds  but when it takes long time  it takes long time but then plugins loading page is faster then the docker

 

[comet424]

i also wondered maybe does unraid optimized to read nvmes   as i found when i had nvme gen4x4   i couldnt get 5000mbs  or close to it in the disk speed test..  i only got 2500  same with my 3500   so i was also wondering maybe docker and unraid is limited to 2500mbs  and cant handle the nvmes?   like windows cant   least not 10  maybe 11

[Squid]

I don't have a PCIe gen 4 on my server, but I can easily hit close to the bandwidth of my Gen 3's.  It goes without saying though that if you don't have a Gen4 slot on your server then you won't get Gen4 speeds.

 

On my Windows box (with a Gen4 SSD and a Gen4 slot) I've got no problems hitting the rated speed of the SSD (WD Black)

[me114]

comet424

 

New to the forum, so not sure if this has been discussed a lot but my NAS always has cpu problems with linux.

I've had problems with 1 or two cores stuck at 100% usage, tracked it to usb auto suspend. 

 

usbcore.autosuspend=-1 disables autosuspend and after reboot its back to normal.