Bug ID 520817: Overcommitment of Hypervisor Resources is not Advised

Last Modified: Apr 10, 2019

Bug Tracker

Affected Product:  See more info
BIG-IP AAM(all modules)

Known Affected Versions:
12.1.0, 12.1.0 HF1, 12.1.0 HF2, 12.1.1, 12.1.1 HF1, 12.1.1 HF2, 12.1.2, 12.1.2 HF1, 12.1.2 HF2, 12.1.3, 12.1.3.1, 12.1.3.2, 12.1.3.3, 12.1.3.4, 12.1.3.5, 12.1.3.6, 12.1.3.7, 12.1.4, 12.1.4.1

Fixed In:
12.0.0

Opened: Apr 30, 2015
Severity: 3-Major

Symptoms

In a VM running VE where datastor is active, boot up can result in a system that is nearly unusable due to an excessively slow system. In some cases, the customer may see "softlockup" errors on their console and in their kernel message log.

Impact

An excessively slow system.

Conditions

If the hypervisor is configured to allow for an overcommitment of resources, what appears to be real memory to the VM could in fact be virtualized memory. If sufficient amounts of memory exist, but more than one VM is running, it is possible that overcommitment may require the real memory to be "moved" from one VM to another, again causing excessive slowness in booting.

Workaround

Don't over commit your hypervisor resources. TMOS is designed to be an embedded system, not a time-share process, so its resource requirements are much more stringent than most VM instances.

Fix Information

The maximum size of the datastor page cache has been capped at about 10 Gigabytes so as to mitigate the risk of this event occurring.

Behavior Change