Jump to content

anyway to improve docker speed?


Recommended Posts

Posted

hi i had to replace my Sata  Cache 1tb  as it was having CRC errors and i experienced slowness of the Docker when you click it..  and slowness of the VMS..  so i decided to go with a NVME 3500/3000  SSD   i get those speeds in Windows close to it..  and in unraid  the diskspeed test says i get 2500 roughly..  so i dunno why its slow there.

 

but i finding  1.. i have had to reinstall docker 3 times in the past 2 weeks it keeps poping up saying my docker could corrupt from an older beta.. so delete and re install.. im currently doing it again right now..  since i trying to fix the speed..

 

2..  i have set the notifications never to check for updates.. and i have the ca update to do it once a day.

 

so when i load  up docker it be slow  takes like 10 seconds. or so...    but after that..  i go to any other tab.. and go back to docker its instantly there...

its like if it was spinning up drives  to access.. but no all the appdata, docker,system,iso are all saved on the SSD cahce... i dont use the cache drive for file transfers. as i dont waant to save the life of the SSD  as  the Sata was less then 6 months old i RMA and was 88% life left.. so i only using for those 4 directories

is there anything else i can check why dockers so slow... 

i also made sure no more errors anymore with Fix Commons... those test with no errors..

and it be the rest of the day to reinstall the 26 Docker items to reinstall.. but i only have 1-3 usually running at one time anyways..

 

but is there anything else i can check why its slow.. shouldnt it be lighting fast with the nvme?

 

  • 1 year later...
Posted (edited)

so i still struggling with Docker speed.. where it can be like 2 3 seconds to 10 seconds to 1min  to populate

also i find that the Transmission  Icon goes away as an orphane but it has come back once i shut down a VMetc

here is a video i made 

here some Diganostics i took also

 

the Docker  acts like a Spinner Drive when put to sleep    takes time to power up..  and get working...    

 

would it be better to set the appdata cache only to  /mnt/cache/appdata   and not   /mnt/user/appdata    this way it wont go to the array 

i not sure?

 

my video of docker never loading.. when i do it takes about 1 min..  i took this  on aug 9

https://1drv.ms/u/s!AkZ74ugB21kngYMnucOZA60pvyfm6w?e=WM2LLm

 

 

so i not sure whats causing it..  the orphane pic is transmission isnt even running but i noticed  if i shut off a VM  the image usually comes back and then Docker will speed up  but it will go back to sleep later

 

currently running on a Sata SSD.. but this happens on my NVME 3500/3000   it doesnt matter what it does its always sleeping  and takes a long time to run but vms always populate before the Dockers.. and in dockers i only running maybe 2-3 apps always   is it a memory thing then?

 

and should you put your VMS on a different SSD then your appdata?  would that help things

i run like 4-5 VMS but they basiclly idlying

 

 

transmissions.PNG

tower-diagnostics-20210809-1045.zip tower-diagnostics-20210812-1033.zip tower-diagnostics-20210812-1020.zip

Edited by comet424
Posted
3 minutes ago, comet424 said:

would it be better to set the appdata cache only to  /mnt/cache/appdata   and not   /mnt/user/appdata    this way it wont go to the array 

cache-prefer will put it on cache unless cache doesn't have enough space. cache-only will fail if cache runs out of space.

 

Why did any of it go to the array? Did you fill cache? You haven't set a Minimum Free on any of your pools.

 

It looks like most of your disks are very full, possibly some of the various problems you have had on various threads is related to that.

 

Posted (edited)

@trurl i have it set cache only   i dont have any cache on the array

 

but 

/mnt/cache    <=== would read the Cache Only

/mnt/user      <==== would read cache and read the array shares

i basiclly commenting on the one guys findings

that setting your Docker appdata    to /mnt/cache/appdata     and not to /mnt/user/appdata     that its possibly acting slower  pointing to user not to cache 

is basiclly what i was trying to get at

 

i have no  appdata/vms/docker/system  they all on the cache   nothing on the array

 

so my cache drive set to Cache only  and has 126gb free onit..  it doesnt really change unless plex gets bigger.. as i dont do any caching

its just vms,docker,appdata,system    so thats only on the cache 

 

download drive  is just a Spinner Cache drive  just  so i can download from transmission or all my Windows Downloads  go directory to this spinner drive so the array stays off

 

lancahce pool drive is just the drive for downloading  Epic Games  as i have to re install the games.

 

so i reported this issue back in 2020  but never had any solution..  where you can see the Docker Video so slow  loading  but VMs page will load faster.

 

i picked up a 12TB drive yesterday so i preclearing it  so thats going to take a week..   and i not sure if you can take the array offline  without messing up the Pre clear progress.,.. i not sure if it stops  if you stop an array..

 

and Fix Common Issues  doesnt find anything

so what what should i look for?

 

so recap after reading all my dislexia writting

non of the 3 pools save to the array.. they all cache only

the appadata,docker,system,vms all on the cache drive

 

i was asking  if its possible  slow

because vms docker   points to /mnt/user/appdata      and maybe i need to change to /mnt/cache/appdata

as another user was finding slower speeds this way too  

and figured it was reading the array when you point to /mnt/user  instead of /mnt/cache

 

hopefully you can understand i know with my dislexia  it sounds ok in my head not always as you read it

 

also why i asked should VMS  and the docker/appdata/system  be on 2 different SSDS too

Edited by comet424
Posted (edited)

but ill check with krusader to make sure none of those 4  folders be on the array but it shouldnt be 

 

i know i may have mixed up words

 

i see i had the cache only set to prefer.. i was doing a NVME transfer   to send in for RMA

and i set to Perfer when i was using Mover

 

as there is no option when  VMS appdata etc  is on the ARRAY

but you set it to CACHE ONLY  that mover will move it from the array to cache..     mover will only move it when you set to PERFER   so thats why its set there

but i should have set things to ONLY after it was done..  it be nice  MOVER would move the Folders from the Array when you set things to CACHE ONLY  so i double checking things with krusdader

 

but i been having this slowness problems for over a year now  even when all the settings were set to Cache only and nothing on the array

Edited by comet424
Posted

According to diagnostics, appdata is cache-prefer, and it has some files on disk5.

 

You can see how much of each disk is used by each user share by going to User Shares and clicking Compute... for the share.

 

 

 

 

Posted (edited)

@trurl  ah ok ya  thats where i mentioned i had to set to cache perfer  because cache only doesnt work when u wanna use the mover  when  you switch  ssd's

 

ok so i did that.. appdata takes forever..  is the problem Docker is slow

due to Plex running and or transmission/deluge

i have noticed  plex takes forever?

should it be 

appdata on its own ssd

vms  on its own ssd

plex docker on its own ssd

 

would that speed it up? here is the diaganostic

 

when i click the compute  its fast for vms,system  but took 30 seconds to do appdata

 

so i was wondering  should we be doing  seperate  SSD to to speed up the system..  least for Plext 

 

as it doesnt matter if its on a Sata SSD or a NVME 3500/3000 or a 7000/5000  nvmes  it was still slow as molasses

so thats why i was also thinking its unraid the bottle neck too maybe when i gone to the faster nvmes

 

 

tower-diagnostics-20210812-1142.zip

Edited by comet424
Posted

ah does my settings look better again?    and docker takes 18 seconds to load counting the second after i press the docker tab

so nothing improved the speed

 

is there a way to run a diaganostic.. to see when you press Docker..  to see what is taking sooo long to load..

 

like  a ping

if you were to load up in windows app say onedrive  it tell ya it took 2.5 seconds

loading IE  took 4 seconds

 

is there a program/diagnostic that records   the load times  when you click Docker  so i could see whats taking sooooooooooo long

as sometimes Docker takes 30-1min to load  prior to your findings some was on the array..

 

is there anything i can record in the terminal... like you have that grep program  i not fluent with linux stuff.. but is there something that can show slowiness so i can fix it ?

 

and still Transmission  os orphaned  and doesnt work even though  its running...    when you click webugui.. its working just not downloading.. 

so kinda frustrating and i dont wanna shut off array or reboot as i doing a 12tb preclear.... so i dont wanna jeopardize it

 

but if there is a plugin to show load times of docker  what would it be called

 

 

Posted

so i turned off Plex..

if it click Docker  still takes around 18 -20 seconds

 

if i click  Dashboard  and the Docker page sections also takes 18-20 seconds to load

 

and the only 2 dockers i have running is 

lan-cache 

krusader

 

does it mean anything?

Posted (edited)

so i skimmed over it..  none helped..  they talked about Avast anti virus..  but i dont use it.. and they talk about gui.. but Docker is the only slow thing

 

its like you leave it idle for a few hours... and when you load up  unraid page.. you can click  VMS it takes a few seconds..  but  it loads.. you click docker..  and she taking like 20 seconds to load..   its like your spinner drivers are spun down and you gotta wait for it to spin up..    now i did try changing  some settings like making sure the cache drive folders like appdata system vms  all said Only..   i figure maybe if set to Perfer  maybe thats causing Docker to be slow it might be reading the array waiting for a disk to spin up.. not 100% sure

also the Transmission  Icon came back..  and no longer and orphane image..  and once that happened   Docker loads in 1 second..  but it took a day after me setting vms  appdata etc  to Cache Only

 

also i changed Cpu scalling to On Demand...  i just guessing..  like that other guy having similar issues as me..

so i going to let it go idle  till tommorow and see if  she still slow

 

i did clean up and deleted alot of junk  so i recovered 2TB.. and i finished cycle 1 of 3  after 51 hours for the 12TB.. .  and if i needed to hit  stop array..  does it stop preclear? i not touching much as i dont wanna put the cycles to waste.. so for now i just letting this idle and see if docker help

 

but what i skimmed from the FAQ didnt help my case 😞  

will see what tomorrow brings

 

 

also read the docker faq  about pinning but i cant do that not without a reboot  and dont wanna mess up my preclearing  so ill try those tests in a week

i have like transmission pinned on a 3rd cpu and thread..    and i used to have cpu isolation  but i was experiencing issues so i turned them  off..  so not sure if i cpu isolate if it help but cant do that  need reboot  and cant do that yet

Edited by comet424

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...