I/O Commands are Hanging on Virtual Machines when Using a Large Number of Physical Disks on Oracle VM
(Doc ID 2094105.1)
Last updated on MAY 12, 2021
Applies to:Oracle Cloud Infrastructure - Version N/A and later
Oracle VM - Version 3.0.1 and later
When several VM guests are being provisioned on an Oracle VM Server, and that some of the VM guests are making use of a large number of Physical Disks, commands that attempt to access these physical disks may hang VM guests or completely hang the Oracle VM Server.
Stacks such as the following one may be printed when attempting to access these disks from the VM guests:
Running the same kind of I/O commands such as fdisk against the /dev/mapper/ entries on the VM Server do succeed whereas they fail on the VM Guests for which this /dev/mapper/ entry exists.
To view full details, sign in with your My Oracle Support account.
Don't have a My Oracle Support account? Click to get started!
In this Document
|1. Tune the xen boot parameter gnttab_max_nr_frames.|
|2. Limit the max number of queues on dom0 side (This only requires reboots of the guests - This only applies to UEK)|
|3. limit the max number of queues on domU side (This only requires reboots of guests)|
|OVM 3.4+ / PCA 2.3.1+|