Feature Request: Faster or Local OS load for (big) images #33

Open
opened 2024-04-11 19:43:05 +00:00 by mik-tf · 5 comments
Owner

Situation

When you deploy to a node that had recent deployment, it takes a long time for the OS to be downloaded. Sometimes, it can take up to 20 minutes. When you know this, it's OK, you can just wait. But the everyday user will think something is off.

Proposition

  • we should not have more than e.g. 40 images
  • they should all be build in a consistent manner and build scripts on a repo e.g. buildah and then flist
  • there is one master tfhub, all the slaves sync from
  • we need to add all tfhubs in the flist and make sure ZERO-OS choses the closest one
  • mention very well on the solutions that these are delivered as is and just an example

Reference

Nelson feedback during the community call.

@despiegk as discussed.

## Situation When you deploy to a node that had recent deployment, it takes a long time for the OS to be downloaded. Sometimes, it can take up to 20 minutes. When you know this, it's OK, you can just wait. But the everyday user will think something is off. ## Proposition - we should not have more than e.g. 40 images - they should all be build in a consistent manner and build scripts on a repo e.g. buildah and then flist - there is one master tfhub, all the slaves sync from - we need to add all tfhubs in the flist and make sure ZERO-OS choses the closest one - mention very well on the solutions that these are delivered as is and just an example ## Reference Nelson feedback during the community call. @despiegk as discussed.
mik-tf added the
Story
label 2024-04-11 19:43:05 +00:00
mik-tf added this to the (deleted) project 2024-04-11 19:43:05 +00:00
Owner

please estimate work and put deadline, assign people

please estimate work and put deadline, assign people
Member

won't be before 7th of may

won't be before 7th of may
sabrinasadik was assigned by thabeta 2024-05-07 11:26:12 +00:00
Member

Reassigned to Sabrina given her team is already responsible for the images and distribution of the hub

Reassigned to Sabrina given her team is already responsible for the images and distribution of the hub
despiegk modified the project from (deleted) to tfgrid_3_14 2024-05-22 07:56:25 +00:00
despiegk removed this from the tfgrid_3_14 project 2024-05-22 07:57:23 +00:00
despiegk added this to the tfgrid_3_16 project 2024-05-22 07:57:45 +00:00

Hi @mik-tf ,
How to reproduce this issue

When you deploy to a node that had recent deployment, it takes a long time for the OS to be downloaded. Sometimes, it can take up to 20 minutes

So, this issue happened when we spawn a new VM and the OS image is not in the node yet?

I tried to reproduce it:

  1. create a new zos node

  2. create a VM on that node from this dasboard
    telegram-cloud-photo-size-5-6165477286745324999-y

  3. it only took 1-3 mins for me from Indonesia/Asia.

Or do you mean the slowness happens when we login to the newly deployed VM

Hi @mik-tf , How to reproduce this issue > When you deploy to a node that had recent deployment, it takes a long time for the OS to be downloaded. Sometimes, it can take up to 20 minutes So, this issue happened when we spawn a new VM and the OS image is not in the node yet? I tried to reproduce it: 1. create a new zos node 2. create a VM on that node from this dasboard ![telegram-cloud-photo-size-5-6165477286745324999-y](/attachments/e8dfa5a7-2f1e-4ece-8a6e-f74486778365) 3. it only took 1-3 mins for me from Indonesia/Asia. Or do you mean the slowness happens when we login to the newly deployed VM

OK, i can reproduce it with nixos image.

I got timeout after 10 mins

OK, i can reproduce it with `nixos` image. I got timeout after 10 mins
Sign in to join this conversation.
No Milestone
No project
No Assignees
4 Participants
Notifications
Due Date
The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Dependencies

No dependencies set.

Reference: tfgrid/circle_engineering#33
No description provided.