How to clean up /dev/shm

[gmcwilliams]

Hi everyone,

Been receiving alerts for a little bit around tmpfs /dev/shm being at 99% - how should we clean up this directory? What is this directory used for?

tmpfs /dev/shm

Thanks!

[BPry]

@gmcwilliams,

Restart the firewall. /dev/shm is used on the firewall the same as it is on any other Linux system, it's a temporary file store that uses memory instead of disk like /tmp does. Outside of restarting the firewall I'm not sure there's a way exposed to actually do anything about it outside of a reload. 

[gmcwilliams]

Does 10.0.* cause this to fill up faster?

EDIT: Is there anything we should worry about if this folder fills up? I just don't want an unexpected crash

[Jatin.Singh]

Typically this is cleared on reboot but after upgrading to 10.0.6 its failed to clear the space on system reboot.

I have looked at the other drives on the PAN are there seems to be no capacity issues other then the tmpfs /dev/shm location at 97%.

Can yoU confirm whether this is expected on later versions 10.0.x +?

[Jatin.Singh]

Where you able to to clean it, where you able to get to know why it is high.

We are having the similar issue

[RSumalal]

We are also seeing same behavior on VM-300 Palos after upgrade to 10.0.5, the tmpfs (/dev/shm)  is on 97%. Do we have any workaround or solution for this?

[gmcwilliams]

I am not sure why it gets so high, but it clears up on reboot. It does not seem to cause any issues while maxing out but it's probably safe to do a monthly reboot or whenever Palo releases a new update

[gmcwilliams · Accepted solution]

The workaround I have is that the Palo Alto needs to be rebooted. While we haven't run into any issues with it at 97% I would say it's safe to do a monthly reboot to clean up the directory.

[Firewall_Guy]

This is what we received from TAC:

When auto refresh for dashboard and content is happening very frequently (every 1-2 min) and continuously, some intermittent shm memory growth is still expected though but it should get released gradually and no big build-up should be observed and /dev/shm should not be filled up. But /dev/shm partition disk space builds-up .Also, expected to see Not enough space to load conent to SHM" errors in the system log.

This issue is already identified and reported on 10.0.0 over VM platforms - Engineering team comments that this fix requires some architectural changes on the report generation part of PanOS. Currently the change has been marked as a feature enhancement and is targeted to be addressed with PanOS 11.0 version or 10.2.0. The tentative target fix version depends on QA validation tests; which usually are completed prior to a week or two of the software release.