Got the connect problem after install the Ceph via Portal

[parker0909]

Hi All,

My name is Parker , i am facing the problem related to ceph.I have tried to install the ceph via Proxmox VE portal, i got the disconnect message during the installation status andi unable to do the configuration. After i refresh the page, it seem the node ceph cant connected to the existing ceph. May i know any idea we can to fix the ceph problem?

I have attached the print screen to show the node 2 - Ceph init node and node 3 should be the node have ceph problem.

Thank you.

 

[parker0909]

I have attached the print screen i got the "got timeout" message during the ceph installation process.

Parker

 

[parker0909]

Hi All,

The good news all node ceph should be normal now, but we facing another question when i try to create the OSD with DB disk. I got below message.


'/dev/sde' is smaller than requested size '800156322201' bytes


May i know how to calculate the smaller than requested size for DB disk? Thank you.

Parker

 

[Alwin]

The DB partition size is 10% of the data disk, when a separate DB/WAL is selected.

 

[parker0909]

The DB partition size is 10% of the data disk, when a separate DB/WAL is selected.

Thank you. May i know it is able to share with different OSD .For example we have three OSD (8TB SATA),so we can share one 800GB SSD for DB disk. May i know it is good idea for DB disk setting?

Parker

 

[Alwin]

Thank you. May i know it is able to share with different OSD .For example we have three OSD (8TB SATA),so we can share one 800GB SSD for DB disk. May i know it is good idea for DB disk setting?

Yes.

 

[parker0909]

Yes.

I got the problem it seem OSD cant share the same SSD DB disk with not enough free space. it meaning should be i need arrange 3 800GB ssd disk for db disk if we have three 8TB OSD. May i know it is correct? Thank you.

3 OSD
3 x 8TB SATA disk
3 x 800GB SSD for DB cache

Error when i try to create another 8TB OSD
vg 'ceph-527c49e7-0546-403c-8606-8b88cff8c0e5' has not enough free space

Parker

 

[Alwin]

I got the problem it seem OSD cant share the same SSD DB disk with not enough free space. it meaning should be i need arrange 3 800GB ssd disk for db disk if we have three 8TB OSD. May i know it is correct? Thank you.

As said, the DB/WAL partition/LV will be 10% of the data disks size. So 4 TB data disk, will result in ~400GB DB/WAL partition/LV. If the partitions geht to big, then you will need to set it in the ceph.conf.
https://forum.proxmox.com/threads/where-can-i-tune-journal-size-of-ceph-bluestore.44000/#post-212147