Ceph Not install In node

[Amol.pande]

Dear Support

Good day,

I am planning on using Proxmox in my environment. I have also created a cluster by installing 3 nodes. And now I want to install Safe, so it is not getting installed. Means is not updating. I have also disabled enterprise. But still same issue.
Will any of you guide me for this help? Because, we have to setup this. After demo success.

Thanks In Advanced
AMol

 

[UdoB]

I have also disabled enterprise.

No, you have not - the first screenshot ("Unauthorized") tells this. Disable it in the Web Gui.

Make sure to enable the "download.proxmox...../debian/ceph-reef" "no-subscription" repository via the Gui. Then try again...

 

[Amol.pande]

No, you have not - the first screenshot ("Unauthorized") tells this. Disable it in the Web Gui.

Make sure to enable the "download.proxmox...../debian/ceph-reef" "no-subscription" repository via the Gui. Then try again...

Dear Sir,

i check can you see this snap shot please

 

[UdoB]

Yeah. But now you disabled the Debian repositories - probably for debugging? Debian is essential for PVE.

You need to re-enable all three of them.

Then run apt update && apt -d -y full-upgrade in a terminal and post the full output (if something fails).

(It would be nice to post it as text, not as a picture. Please use [code]...[/code]-tags for this.)

 

[Amol.pande]

Yeah. But now you disabled the Debian repositories - probably for debugging? Debian is essential for PVE.

You need to re-enable all three of them.

Then run apt update && apt -d -y full-upgrade in a terminal and post the full output (if something fails).

(It would be nice to post it as text, not as a picture. Please use [code]...[/code]-tags for this.)

Dear UDOB

Thanks A lot for support,

now i am able to fetching update from my firewall block this update

I had one more question.
When I was installing the server, I had created 200 + 200 GB space.
When I came near the drive section. Then I selected ext 4.
But it was also showing raid 0. So can I raid 1/5 from that if I have 2 drives?
Or all drives have to be kept as ext 4 at the time of installation.

Thanks
Amol

 

[UdoB]

now i am able to fetching update from my firewall block this update

Great; you're welcome. Click on the "Like" button of the helpful post ;-)

When I was installing the server, I had created 200 + 200 GB space.
When I came near the drive section. Then I selected ext 4.
But it was also showing raid 0. So can I raid 1/5 from that if I have 2 drives?
Or all drives have to be kept as ext 4 at the time of installation.

That's difficult to understand. "200 + 200" for... what? For base OS and for LVM? I do not install from scratch so often, so I do have the setup sequence and option in my mind, you know ;-)

Changing the partition layout and sizing after completing the installation is... difficult. If you are still in the experimenting phase I would recommend to reinstall and select the appropriate layout.

Remove a node (https://pve.proxmox.com/pve-docs/pve-admin-guide.html#_remove_a_cluster_node) first and add it back after fresh installation. I am not sure about left over artifacts like key fingerprints; the way to avoid this topic is to remove "pve1" and add the same hardware as "pve4".

I can recommend to avoid Raid0 as this means zero redundancy. Use Raid1 = Mirroring. Personally I am a fan of ZFS, all of my systems use this...

Good luck.

 

[Amol.pande]

Great; you're welcome. Click on the "Like" button of the helpful post ;-)



That's difficult to understand. "200 + 200" for... what? For base OS and for LVM? I do not install from scratch so often, so I do have the setup sequence and option in my mind, you know ;-)

Changing the partition layout and sizing after completing the installation is... difficult. If you are still in the experimenting phase I would recommend to reinstall and select the appropriate layout.

Remove a node (https://pve.proxmox.com/pve-docs/pve-admin-guide.html#_remove_a_cluster_node) first and add it back after fresh installation. I am not sure about left over artifacts like key fingerprints; the way to avoid this topic is to remove "pve1" and add the same hardware as "pve4".

I can recommend to avoid Raid0 as this means zero redundancy. Use Raid1 = Mirroring. Personally I am a fan of ZFS, all of my systems use this...

Good luck.

Dear UdoB,

Greetings!!

Actually I tried this a lot for ZFS. But it is not visible. I have given that snap. When I go to configure?
actually I am doing this setup in hyper v. And in that I have earned two virtual disks worth Rs. 200 for testing.
When I start the installation from the beginning, I get the mirror option. But I installed from ext4 and played a video from OSD.
I see some snaps, you will see once.

you have any full installation step from start to all configuration if possible

Please

Thanks
Amol

 

[UdoB]

you have any full installation step from start to all configuration if possible

Well..., no. I can not repeat an installation guide like

• https://pve.proxmox.com/pve-docs/pve-admin-guide.html#chapter_installation
• https://pve.proxmox.com/wiki/Installation

Screenshot 1+2 are okay. Screenshot 3 does not fit to those first two. In the background I can see you have actually configured LVM, not ZFS.

To get rid of LVM the easiest way is just to reinstall ;-)

If you want to add an addtitional ZFS pool while keeping LVM you can add two (or any number) more fresh disks, and then screenshot 3 will list those two as "new unused disks".

Good luck!

 

[Amol.pande]

Dear UdoB Sir,

Thanks For Guide and support,

Now i create a new fresh installation and create a 100 GB for OS 200+200 For ZFS pool Now i have a another why the need for OSD meanse what is the role of OSD for Ceph

Thanks
Amol

 

[UdoB]

Now i have a another why the need for OSD meanse what is the role of OSD for Ceph

ZFS is local only, Ceph is distributed only. (And a single Object-Storage-Daemon handles one physical disk.)

Both have different capabilities: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#_storage_types

I am sure you can find some primers "out there"... ;-)

 

[Amol.pande]

Dear UdoB Sir,

Greetings!!!

Many of my points have become clear from you. Thank you very much for sharing nice knowledge.
Mostly my 3 nodes are ok. As far as I think. I have given some snap. Can you watch it once, whenever you get time?



Can i used OSD storage also for VM installation Right?

And what extra needs to be done in this? Can you tell some points? As soon as this is finalized, I am going to show this work to my management. And then we'll go to Enterprise.

Thanks you so much Sir ones again

Amol

In here can i select both storage data in EFI Disk And TPM

 

[Amol.pande]

Dear Sir,
I am trying to assign 20Gb Ram but it is using 75% why this using high means i am using free or using hyper v

 

[gfngfn256]

Dear Sir,
I am trying to assign 20Gb Ram but it is using 75% why this using high means i am using free or using hyper v


View attachment 68587

Don't understand your question, but that image is weird; the Intel E5-2640V3 CPU should have 8 cores as intel docs state here, but you are only showing 7. Don't know what's up (maybe its virtualized somehow) - but it doesn't look right.

 

[Amol.pande]

Don't understand your question, but that image is weird; the Intel E5-2640V3 CPU should have 8 cores as intel docs state here, but you are only showing 7. Don't know what's up (maybe its virtualized somehow) - but it doesn't look right.

Dear Sir,

This was my question. I have given 30 GB RAM for each node. And for vm I have given 20Gb ram.
But if I use less processor or RAM then the installation takes about 5 to 6 hours.

So why does it take so much time? Is this a question?

And I am using un subscribe. Therefore hardware utilization is increasing.? Had to ask this?

see the task logs


so processor can i use

Hyper-v

Memory using high 90% +


Can't we keep it stable like this? Meaning I am installing only one VM. According to the rest of the VM, everything becomes proper in EXSI. Nutanix also runs well in ISO community, in demo. But just installing vm in proxmox takes a lot of time and even shutdown or reboot doesn't take a lot of time.

Thanks
Amol

 

[UdoB]

Can you watch it once, whenever you get time?

Your setup is both too restricted and too complicated.

Virtualizing a complete datacenter with three nodes plus Ceph needs a lot of details "to fit". Sorry, I can not point to a specific details.

I have given 30 GB RAM for each node. And for vm I have given 20Gb ram.

30 GB per per virtual node (and giving 20 GB too a VM) maybe far too less. The Node needs some GB for its basic functions. Every single OSD, every MGR and every MON needs a Gigabyte or so. (I didn't measure.)

One basic rule I've learned: yes, you can overcommit CPU a lot. No, you can NOT overcommit RAM. (Only a handful percent, perhaps...)

Give Win-10 "only" 10 GB and retry.

Can i used OSD storage also for VM installation Right?

Yes, sure. Of course not "direct" but via usage of a block device, managed by PVE and hidden "below" the GUI.

I am trying to assign 20Gb Ram but it is using 75% why this using high means i am using free or using hyper v

Again difficult to understand. "High Ram usage" has also been discussed several times here in the forum. Most times it was a misunderstanding of what "free RAM" means (on the node and inside a guest) and if Windows is involved, the fact that Windows tends to grab the whole amount of RAM immediately. (Depending on the actual configuration, ballooning and/or PCI pass through.)


Perhaps a virtualized datacenter is not a good idea for a beginner - you are putting one complex system (PVE) onto another complex system (your Hxper-V) and then add another, third complex system (Ceph) on top of that. All systems interfere. Debugging is not easy, sorry...

Good luck

 

[Amol.pande]

Your setup is both too restricted and too complicated.

Virtualizing a complete datacenter with three nodes plus Ceph needs a lot of details "to fit". Sorry, I can not point to a specific details.

30 GB per per virtual node (and giving 20 GB too a VM) maybe far too less. The Node needs some GB for its basic functions. Every single OSD, every MGR and every MON needs a Gigabyte or so. (I didn't measure.)

One basic rule I've learned: yes, you can overcommit CPU a lot. No, you can NOT overcommit RAM. (Only a handful percent, perhaps...)

Give Win-10 "only" 10 GB and retry.


Yes, sure. Of course not "direct" but via usage of a block device, managed by PVE and hidden "below" the GUI.


Again difficult to understand. "High Ram usage" has also been discussed several times here in the forum. Most times it was a misunderstanding of what "free RAM" means (on the node and inside a guest) and if Windows is involved, the fact that Windows tends to grab the whole amount of RAM immediately. (Depending on the actual configuration, ballooning and/or PCI pass through.)


Perhaps a virtualized datacenter is not a good idea for a beginner - you are putting one complex system (PVE) onto another complex system (your Hxper-V) and then add another, third complex system (Ceph) on top of that. All systems interfere. Debugging is not easy, sorry...

Good luck

Dear UdoB,

Good day,

Thanks for replying. thank q so much

Thanks
Amol