GitHub user SviridoffA added a comment to the discussion: Cloudstack Migrate VM Query
There are two HA mechanisms: VM HA and Host HA. VM HA monitors the health of virtual machines and restarts them if the power report on the host indicates a failure, or if the host they were running on switches to the "down" state. In the first case (power report failure), the restart usually happens after about three minutes. In the second case (host down), it takes roughly five minutes — after CloudStack completes all host checks and confirms that the host is truly unreachable. Host HA mainly handles fencing of problematic nodes. Which mechanisms combination will work best depends entirely on your infrastructure — especially on the type of storage you are using (NFS always wins). Below I’ve added a link to a post that describes the features and behavior of these mechanisms in more detail. Personally, I use Linstor. Unfortunately, none of the current mechanisms are acceptable to me, so I built my own custom automation based on monitoring and special HA virtual machines. However, it turned out to be quite complex, so now I’m planning to switch to the more native VM HA combined with my own custom fencing automation. What will work best for you is ultimately a matter of practice and testing. Some people use only VM HA, some use both mechanisms, and others create their own custom solutions. GitHub link: https://github.com/apache/cloudstack/discussions/12629#discussioncomment-15780196 ---- This is an automatically sent email for [email protected]. To unsubscribe, please send an email to: [email protected]
