In certain configurations, ClamAV may use excessive storage on a VM, possibly filling the disk space and making the environment unstable. It may use up all available storage on the /var/vcap/data/ volume. This applies in particular to the nfs_server VM and the diego_cell VMs but, it can affect any VM on which ClamAV is configured with both on-access and scheduled scans enabled at the same time.
This issue occurs because ClamAV holds on to deleted file handles in /var/vcap/data/tmp. The problem arises when both on access and scheduled are set to yes (on_access: yes and scheduled: yes)which causes a race condition that does not release the deleted file handles.
The workaround is to set property for on_access: to no or set scheduled: no.
For the workaround example below, we have set on_access: no.
releases:
- { name: fim, version: "1.1.2" }
- { name: clamav, version: "1.1.6" }
addons:
- name: fim
jobs:
- name: fim
release: fim
properties: {}
- name: clamav
jobs:
- name: clamav
release: clamav
properties:
clamav:
on_access: no
scheduled: yes
database_mirror: 10.147.50.20
The fix for this issue will be available in the next release.