Welcome Guest. Please Login or Register  
Client Area  |  Softaculous Cloud  |  Calendar  |  Quick Links  |  
   


You are here: Index > Virtualizor - Virtual Server Control Panel > General Support > Topic : Memory Overcommitment and Backup-Induced VM Downtime



Threaded Mode | Print  

 Memory Overcommitment and Backup-Induced VM Downtime (4 Replies, Read 30201 times)
DashRDP
Posted on # January 6, 2025, 8:06 pm
Group: Member
Post Group: Newbie
Posts: 2
Status:
We have a server running Virtualizor with three VMs, each configured with an overcommitted RAM allocation of 10 GB. During idle periods, the RAM usage for each VM remains under 1 GB. However, during their respective backup processes, each VM utilizes its full allocated RAM. Post-backup, the RAM usage within the VMs reduces back to approximately 1 GB, but the host node still reports higher RAM usage for the VMs than what is observed inside them.

Attempts to clear the cache and buffer on the host/VMs have not resolved the issue. The main problem occurs when all three VMs initiate their backup processes simultaneously, resulting in excessive memory demand that causes the VMs to become unresponsive or go offline.
IP: --   

Memory Overcommitment and Backup-Induced VM Downtime
jevingala
Posted on # January 7, 2025, 6:15 am | Post: 1
Group: Virtualizor Team
Post Group: Super Member
Posts: 773
Status:
Hi,

Do you see any process still active in background once the backup is finished ?

ps aux |grep -E 'backup|qemu-monitor'

Also ksm service is not active on that server ?

-----------------------
Regards,
Virtualizor Team.
http://virtualizor.com/
IP: --   

Memory Overcommitment and Backup-Induced VM Downtime
DashRDP
Posted on # January 7, 2025, 11:29 am | Post: 2
Group: Member
Post Group: Newbie
Posts: 2
Status:
The issue is not specifically the backup process, but when the VM utilizes the full memory in any way and later the memory usage drops inside the VM, the host node still reserves the memory.

Here VM1 is utilizing under 1GB RAM https://prnt.sc/c89T3EahZw2d
But on the host node, the same VM1's process is utilizing more RAM https://prnt.sc/HI4NpRlVv_4a

And yes KSM is active and running.
IP: --   

Memory Overcommitment and Backup-Induced VM Downtime
emilyjohnson63
Posted on # February 3, 2025, 10:44 am | Post: 3
Group: Member
Post Group: Newbie
Posts: 1
Status:
It sounds like even though the VMs aren't using all 10GB normally, the hypervisor (Virtualizor in this case) is still reserving or allocating that full amount to them, just in case.  When the backups kick in and they do need it, the host is scrambling to find enough RAM to give them all their allocated amounts at the same time, which is why they're crashing.  Clearing cache won't help because it's not about cached data, it's about the reserved RAM.  You've basically over-promised RAM that the host doesn't physically have when everyone needs it at once.  You probably need to either reduce the overcommitted RAM allocation for each VM (maybe 6GB or 8GB instead of 10GB) or stagger the backups so they don't all run at the same time.  That way the host has enough RAM to go around.  It might also be worth checking if Virtualizor has any settings related to memory ballooning or dynamic memory allocation – those might help, but no guarantees.
IP: --   

Memory Overcommitment and Backup-Induced VM Downtime
Bevisflyn
Posted on # February 28, 2025, 6:28 am | Post: 4
Group: Member
Post Group: Newbie
Posts: 3
Status:
Can you provide more information about the configuration of each VM? Specifically the number of CPUs, the type of OS, and the applications running on each VM? This may help determine what is causing the high memory usage during the backup.
IP: --   

« Previous    Next »

Threaded Mode | Print  



Jump To :


Users viewing this topic
2 guests, 0 users.


All times are GMT. The time now is May 9, 2025, 4:10 am.

  Powered By AEF 1.0.8 © 2007-2008 Electron Inc.Queries: 10  |  Page Created In:0.180