https://rancher.com/ logo
Join Slack
Powered by
This message was deleted.
# harvester
a
This message was deleted.
f
Both should work, but I'm not sure which is the best practice for longhorn. There are many things to consider: performance, security, and ease of operation and maintenance. I created an issue on github https://github.com/harvester/harvester/issues/3183 to track this and get advice from longhorn experts
p
Say if you have two 100GB disks on each node. They both mounts as seperate extra disks. Now when I upload an image 125GB, it fails saying “No space left”. How do we handle images larger than a size of a single disk?
f
In this case, it seem that the only option is to increase the size of the hard drive, either by using multiple hard drives for RAID or by using a larger single hard drive
p
Hmm ok. How about creating a virtual machine or a volume larger than 100GB? Have not tried yet but what is going to happen?
f
longhorn replica can't be scheduled because there are no disks have enough space
p
Say I have , 3 nodes. each with 2x 100GB disks
The VM i am creating is 125Gb in size
f
Oh, sorry. I forgot that longhorn storage can be overcommit, since the default storage overcommit setting is 200%. I guest it will success. https://docs.harvesterhci.io/v1.1/advanced/settings#overcommit-config
p
May be I am asking too much , but how about if you need a thick provisioned vm?
f
What do you mean 
a thick provisioned vm
? the example vm with 125G disk?
p
RIght. I mean nonsparse or preallocated disk volume
f
Maybe I will use 
qemu-img
to shrink vm image first
p
The image is a qemu-img converted image and it took down to 125GB from 275GB.
f
The vm is successfully created in favor of storage overcommit.
p
Ok
Thank you. I will test it when I can and report here
154 Views
Open in Slack
PreviousNext
Powered by