node can't deploy a workload even if it covers the required capacity

threefoldtech / zos

Autonomous operating system

https://threefold.io/host/

Apache License 2.0

83 stars 13 forks source link

node can't deploy a workload even if it covers the required capacity #2351

Closed Omarabdul3ziz closed 3 months ago

Omarabdul3ziz commented 3 months ago

for example. node 11 have total_sru=512110190592 and used_sru=480436551680 so it should have free 31673638912 around 29GB but it can't deploy a workload asking for sru more than 22GB.

if fails with this error:

failed to create vm rootfs: not enough space left in pools of this type ssd

it may be because the node splitting its SSD to pools and no pool can cover the requested SSD https://github.com/threefoldtech/zos/blob/02480a938b32294cca7031866262a017c0bae7fb/pkg/storage/storage.go#L741

Omarabdul3ziz commented 3 months ago

after some debugging here what i found:

both grid clients go/ts do ask for zos.storage.pools before it considers the node appropriate to host this workload.
what will happen in the parent issue is, the script asks for sru with only the disk size which the node can cover its capacity. but will deploying the machine required extra sru for rootfs which the node couldn't afford.

reported to edit the script.

it is not an issue, safe to close

Mahmoud-Emad commented 3 months ago

What is the required extra sru for rootfs? 2GB?

Omarabdul3ziz commented 3 months ago

as far as i remember if the provided rootfs is zero, there is an equation that will calculate it from the CU (ram/cpu) or at least yes will be 2GB. @AhmedHanafy725 can clarify more on that