Forum
Welcome, Guest
Username: Password: Remember me

TOPIC:

Ha-lizard on the XCP-NG 8.2 in 2021. Progress of my deployment 4 months 3 weeks ago #2449

  • Salvatore Costantino
  • Salvatore Costantino's Avatar
  • Offline
  • Posts: 696
There is one scenario where conversion of the local SR fails.

Our default installer requires that the block device used for local storage does not span more then one backend disk. Meaning, during the installation of XCP-ng, only a single disk is selected for VM storage. The installer will then split that storage device into a number of partitions, with the local SR being on partition 3. When running our installer, we expect this layout.

If more than a single disk is selected when installing XCP-ng, then the XCP-ng installer will add LVM on top of the multiple selected disks and present than as a single storage device when it creates the local SR. We don't handle this environment and the installer will fail to convert the local SR. Perhaps this is what you have run into.

Please Log in or Create an account to join the conversation.

Ha-lizard on the XCP-NG 8.2 in 2021. Progress of my deployment 4 months 3 weeks ago #2450

Apparently the glusterfs driver alert is a problem with the xcp-ng center, because doing the same procedure on the server screen (or xsconsole) everything works fine.

Please Log in or Create an account to join the conversation.

Ha-lizard on the XCP-NG 8.2 in 2021. Progress of my deployment 4 months 3 weeks ago #2451

Salvatore, in my case I installed xcp-ng on a RAID1 (formed by 2 SAS disks of 146 GB 10krpm) and selected as storage of the VM another RAID1 (formed by 2 SAS disks of 146 GB 15krpm). The HA System has worked well, making the migration of VMs between servers. And I am sure that each RAID has a function because I did the HD withdrawal test. Could this structure cause me problems in the future? Master and Slave are the same. These 2 RAID are formed by the HP server itself. Would it be more recommended to create another RAID configuration on the HP Server to deliver to the XCP-NG installer?

Please Log in or Create an account to join the conversation.

Ha-lizard on the XCP-NG 8.2 in 2021. Progress of my deployment 4 months 3 weeks ago #2452

Definitely something I intend to promote... Have you had any direct correspondence with the development committee yet?

I can imaging you running into a conflict of interest complication because HA-Lizard is essentially a competitor to Xen Orchestra where their paid services are HA options.

Please Log in or Create an account to join the conversation.

Last edit: by Konos Koncepts.

Ha-lizard on the XCP-NG 8.2 in 2021. Progress of my deployment 4 months 3 weeks ago #2453

  • Salvatore Costantino
  • Salvatore Costantino's Avatar
  • Offline
  • Posts: 696
Your environment should work fine with the 2 RAIDs. I have deployed many in the same exact way where RAID1 is used solely for dom0 and RAID2 is used for VM storage.

Please Log in or Create an account to join the conversation.

Ha-lizard on the XCP-NG 8.2 in 2021. Progress of my deployment 4 months 3 weeks ago #2454

  • Salvatore Costantino
  • Salvatore Costantino's Avatar
  • Offline
  • Posts: 696
I have tried to reach out, but no response.

Regarding competition, this project is not a commercial endeavor for me. I give it away for free and support it for free, unless someone wants direct paid support. There is a bit of a misconception about the risks that come with a 2 node cluster as is relates to DRBD. We do not run DRBD in dual primary role, which can be dangerous. We allow only a single host to expose storage and then layer on multiple validations to ensure that a VM disk can never be mounted by more than one host. Perhaps publishing crash scenario test results would alleviate some of the concerns.

I welcome any effort you are willing to put forth as installation, maintenance and upgrades would be far easier if our packages were included in the base XCP repo.

Please Log in or Create an account to join the conversation.