Feature #12860
openadd mmc-utils package to all images
0%
Description
Both Netgate & 3rd party hardware integrators are increasingly using eMMC components.
SATA (& historically SCSI) drives have offered SMART reporting & analysis - however SMART does not work with eMMC drives.
eMMC v5.0 standard (published in 2013) specifies simple access to eMMC lifetime estimates & EOL info via extcsd data:mmc extcsd read /dev/mmcsd0rpmb | egrep -i ^emmc
Fortunately there already exists a FreeBSD port of mmc-utils package which is tiny (40-60kB) and so could be easily added into pfSense images.
Hopefully a simple means of GUI access (eg Diagnostics / eMMC Status) to access these lifetime estimates & EOL info could also be considered.
Updated by Steve Y about 3 years ago
This would be helpful/useful now that ZFS is the new default, and/or for folks who don't realize some packages are "required" or recommended to use an SSD (https://www.netgate.com/supported-pfsense-plus-packages).
Forum thread: https://forum.netgate.com/topic/170128/emmc-write-endurance/
Updated by Jim Pingle about 3 years ago
- Project changed from pfSense to pfSense Packages
- Category changed from S.M.A.R.T. to New Package Request
- Release Notes deleted (
Default)
We already build mmc-utils
for Plus and it can be installed manually from the CLI. Trying to build a GUI around it to make any kind of interpretation of the results is trickier. The values people are interested in are largely estimates and may not bear any relation to the actual situation on the disk. We've already been monitoring its output on a variety of hardware of different ages trying to see if it's a reliable metric. We don't want to present a big scary red flag warning a user if it's unwarranted.
I'm not sure it would need to be built into all base system images, especially CE. We could consider at least enable building it and offering it in the CE package repository as we do for Plus.
Updated by Andrew Almond 4 months ago
Has there been any progress on including mmc-utils
in the base image or as a package?
I've experienced several firewall failures this year, which turned out to be due to the onboard eMMC storage dying. A spot check of some running firewalls has revealed others that are possibly near death.
A long-running thread on the Netgate forums speaks to this issue: https://forum.netgate.com/topic/170128/emmc-write-endurance/
I also recently started a thread that highlights the challenges presented by eMMC storage and the lack of out-of-the-box health monitoring. https://forum.netgate.com/topic/195896/concerns-and-feedback-about-storage-lifetime-wearout-on-netgate-devices
We don't want to present a big scary red flag warning a user if it's unwarranted.
This approach seems contrary to the decision to include the ISC DHCP deprecation message.
I understand not wanting to alarm users unnecessarily, but the cautious approach would be to warn users of potential problems rather than have their firewall unexpectedly die.
Including a default widget that shows the output of mmc extcsd read /dev/mmcsd0rpmb | egrep -i ^emmc
in a text box and putting the value-mapping tables underneath and a link to the relevant documentation would be much better than nothing. Or to be less alarming, perhaps have the command run in the background on schedule and only display a message and/or send an alert when the values reach 90%.