I would like to share some cool hardware that we use, I used to purchase hardware from ebay but some sellers ship to my country badly and they arrive pretty badly damaged on arrival. There are some sellers that I would recommend buying from. This japanese seller. He packs quite well. Most of the hardware nowadays I purchase them from sellers from China as they packed the hardware pretty well. They can also help you do setup and customization on the server e.g setting up IDRAC etc so you save time. Their price is also cheaper. I can introduce you to the seller but they are chinese speaking.
Ebay link to seller

C6420 is a pretty good machine for XCP NG


It supports 2 NVME u2 drives and 4 ssds per node.

Each node supports 1 boss card

2 x Gen 1 , Gen 2 Scalable Xeon

The power consumption per node is also pretty decent

Hi everyone, I'm Eddie from Singapore. I run a small business that build mobile apps and websites for customers and we need a rock solid reliable Infrastructure to house and protect all our customers. I have been using XCP NG since the birth of it and have been early backers of XCP NG .

@olivierlambert is my role model and I'm amazed with the growth of XCP NG under his leadership. In this post I want to share some simple setups that keeps my infrastructure rock solid since 2018.

This setup might not be suitable for your use case but it is pretty reliable and rock solid since day 1 and have survived 2-3 instances of disk failure. Before I start sharing my setup I would like to share what I tried previously and why I decided not to use them.

Previously this was my setup

Advantage : This setup is pretty good initially when there are no issues and you can move VMs easily around with ease. Easy maintenance of servers.

Disadvantages :

• You rely greatly on your switch and san storage. If it fails, all your customers will go crazy

• Storage maintenance is harder.

• Your IOPS is limited to your SAN storage capability and Network speed.

Currently our setup uses the KISS approach. Keep it sweet simple approach
Each Nodes have local storage.

1. We use NVMEs and SSDs without raid and local storage. Each node have 10g/25g
2. We do 2 backups with 3 days retention. 1 Delta backups , 1 CR backup to spare servers. Truenas is a pretty good and reliable storag software. We also setup spare XCP NG servers with 8-16 local storage and replicate them.
3. We use tools to monitor disk health.
4. All our servers are 10-20% utilized.

What happens if a server / disk is failing

With the above setup, below are some easy steps to handle it .

1. Step 1, Spin the replica from Replication.
2. The backup from truenas is an additional backup.

This approach seems better for us as we believe that the infrastructure should be very easy for anyone to do the replacement and repairs. E.G CEPH Cluster and ZFS , NFS shared storage etc all requires an experienced engineer to repair them. With this setup and Xen Orchestra, we an easily restore broken vm/server easily and within mins it will be restored.

Thanks wanted to write that my issues are all fixed after grabbing a fresh copy and rebuild.

@olivierlambert i'm doing the git bisect . Will update the commit which cause this issue from restore backup and CR issue.

@olivierlambert Thanks Corrected. I'm so happy for XCP-NG. I'm one of the early backers in Kickstarter. Hang the shirt up in our office.

Did you managed to solve this Tristis ?

I start to have these VDI Error too fresh Xen 8.2 on continuous replications.

I'm using xoa from sources.

I would like to share some cool hardware that we use, I used to purchase hardware from ebay but some sellers ship to my country badly and they arrive pretty badly damaged on arrival. There are some sellers that I would recommend buying from. This japanese seller. He packs quite well. Most of the hardware nowadays I purchase them from sellers from China as they packed the hardware pretty well. They can also help you do setup and customization on the server e.g setting up IDRAC etc so you save time. Their price is also cheaper. I can introduce you to the seller but they are chinese speaking.
Ebay link to seller

C6420 is a pretty good machine for XCP NG


It supports 2 NVME u2 drives and 4 ssds per node.

Each node supports 1 boss card

2 x Gen 1 , Gen 2 Scalable Xeon

The power consumption per node is also pretty decent

Hi everyone, I'm Eddie from Singapore. I run a small business that build mobile apps and websites for customers and we need a rock solid reliable Infrastructure to house and protect all our customers. I have been using XCP NG since the birth of it and have been early backers of XCP NG .

@olivierlambert is my role model and I'm amazed with the growth of XCP NG under his leadership. In this post I want to share some simple setups that keeps my infrastructure rock solid since 2018.

This setup might not be suitable for your use case but it is pretty reliable and rock solid since day 1 and have survived 2-3 instances of disk failure. Before I start sharing my setup I would like to share what I tried previously and why I decided not to use them.

Previously this was my setup

Advantage : This setup is pretty good initially when there are no issues and you can move VMs easily around with ease. Easy maintenance of servers.

Disadvantages :

• You rely greatly on your switch and san storage. If it fails, all your customers will go crazy

• Storage maintenance is harder.

• Your IOPS is limited to your SAN storage capability and Network speed.

Currently our setup uses the KISS approach. Keep it sweet simple approach
Each Nodes have local storage.

1. We use NVMEs and SSDs without raid and local storage. Each node have 10g/25g
2. We do 2 backups with 3 days retention. 1 Delta backups , 1 CR backup to spare servers. Truenas is a pretty good and reliable storag software. We also setup spare XCP NG servers with 8-16 local storage and replicate them.
3. We use tools to monitor disk health.
4. All our servers are 10-20% utilized.

What happens if a server / disk is failing

With the above setup, below are some easy steps to handle it .

1. Step 1, Spin the replica from Replication.
2. The backup from truenas is an additional backup.

This approach seems better for us as we believe that the infrastructure should be very easy for anyone to do the replacement and repairs. E.G CEPH Cluster and ZFS , NFS shared storage etc all requires an experienced engineer to repair them. With this setup and Xen Orchestra, we an easily restore broken vm/server easily and within mins it will be restored.

Thanks wanted to write that my issues are all fixed after grabbing a fresh copy and rebuild.

@julien-f is there similar codes in restore function. The backup delta seems fixed but the restore showing the same error.

@olivierlambert I tested few weeks back with 19 not working too.

@arsenieciprian I reported this issue awhile back when restore is not working but backing up delta is working. Only recently there are more people reporting this issue during CR and Delta backups.

I'm at commit b1e81d8 which Replication and delta works fine now. I'm trying to find the commit where restore starts working.

@arsenieciprian is your restore working fine ?

@arsenieciprian Bruno it means LTS

I'm using the latest lts which is 18

@olivierlambert i'm doing the git bisect . Will update the commit which cause this issue from restore backup and CR issue.