Can-o-Worms: Pi Shutdown vs Power Off
-
@gloomyandy said in Can-o-Worms: Pi Shutdown vs Power Off:
@bearer If you are going to use a read only pi filesystem for the sbc, where are you going to upload gcode files to? Do you intend to use a 2nd disk (network share?) of some sort?
you get an overlay which is read/write worst case you loose your uploaded files and the os and config remain intact.
-
@gloomyandy said in Can-o-Worms: Pi Shutdown vs Power Off:
@bearer If you are going to use a read only pi filesystem for the sbc, where are you going to upload gcode files to? Do you intend to use a 2nd disk (network share?) of some sort?
You could create a partition just for gcode files that's writable and leave the rootfs read only.
-
The only time I've really seen corruption happen is when you loose power in the middle of a high I/O event. Unless your using a PI to build a NAS that you plan on hammering on, you will probably never see a corruption event.
-
@gtj0 said in Can-o-Worms: Pi Shutdown vs Power Off:
Connect 12V to channel 1 and 5v to channel 2. Set the scope it to trigger on channel 1 falling edge.
Roughly what I was thinking. What voltage do we consider, as the 5V decays, to be the limit?
-
@Phaedrux said in Can-o-Worms: Pi Shutdown vs Power Off:
Worst case scenario for a Pi would be maybe replacing the SD card AND the pi itself.
Huh? What mechanism increases or decreases the odds of damage to a Pi (not the SD, the Pi) if it is powered off before or after a Raspbian shutdown command?
-
@Danal said in Can-o-Worms: Pi Shutdown vs Power Off:
Huh? What mechanism increases or decreases the odds of damage to a Pi (not the SD, the Pi) if it is powered off before or after a Raspbian shutdown command?
Exactly. But I just chose that as an absolute worst case total loss situation on one side of the balance.
-
@Phaedrux said in Can-o-Worms: Pi Shutdown vs Power Off:
Totally agree, and that's where the risk must be weighed. How much time will be wasted in the case of a failure? How can that time be mitigated with backups, etc?
Obviously varies by user and Pi. If for some reason I thought a given Pi would take more than an hour or so to rebuild its SD, I'd do something. I'd be much more likely to schedule a backup than to worry about shutdown. SDs fail. Whether shutdown or not.
-
@Danal said in Can-o-Worms: Pi Shutdown vs Power Off:
SDs fail. Whether shutdown or not.
And power loss can happen even with a UPS. (accidentally unplugging it, battery dying, etc)
-
@Phaedrux said in Can-o-Worms: Pi Shutdown vs Power Off:
@Danal said in Can-o-Worms: Pi Shutdown vs Power Off:
Huh? What mechanism increases or decreases the odds of damage to a Pi (not the SD, the Pi) if it is powered off before or after a Raspbian shutdown command?
Exactly. But I just chose that as an absolute worst case total loss situation on one side of the balance.
Huh again?
It seems we just agreed there is absolutely zero change in the odds of losing or not losing the hardware with shutdown or no shutdown. Therefore, hardware loss events do not enter into this discussion.
Yes? No?
-
@Danal said in Can-o-Worms: Pi Shutdown vs Power Off:
@Phaedrux said in Can-o-Worms: Pi Shutdown vs Power Off:
@Danal said in Can-o-Worms: Pi Shutdown vs Power Off:
Huh? What mechanism increases or decreases the odds of damage to a Pi (not the SD, the Pi) if it is powered off before or after a Raspbian shutdown command?
Exactly. But I just chose that as an absolute worst case total loss situation on one side of the balance.
Huh again?
It seems we just agreed there is absolutely zero change in the odds of losing or not losing the hardware with shutdown or no shutdown. Therefore, hardware loss events do not enter into this discussion.
Yes? No?
All I was saying that even if the worst case was a total write off of your SD card and Pi it's ~50$ for the sake of argument. If something cost you 50$ and an hour of your time, how much effort would you be willing to put into mitigating it. It's just a thought experiment.
-
I guess in the context of using a Pi or similar as the SBC on a Duet3, it's pretty rare that anything is going to be written to the filesystem that's of any significance. Aside from doing a software update or uploading a gcode file, typical usage is pretty much just going to be read operations which aren't going to care about getting interrupted anyway.
The test is still really interesting to know how unlikely a failure is. I imagine that the more likely failure more will be due to SD dying due to swap spaces and temporary files being written... but that's a whole separate topic.While I'd always prefer to do a 'proper' shutdown, I've never felt the need on a Pi, and your test seems to have confirmed that. Though if there was an easy option to do so on the DWC, I'd likely use it.
-
@Danal said in Can-o-Worms: Pi Shutdown vs Power Off:
Therefore, hardware loss events do not enter into this discussion.
Is this the can o worms phase of the discussion? I think I'll bow out now.
-
@Phaedrux said in Can-o-Worms: Pi Shutdown vs Power Off:
total write off of your SD card and Pi
I get the time 'cost' and completely agree with you. I don't get the "and Pi" part.
-
@Danal said in Can-o-Worms: Pi Shutdown vs Power Off:
@gtj0 said in Can-o-Worms: Pi Shutdown vs Power Off:
Connect 12V to channel 1 and 5v to channel 2. Set the scope it to trigger on channel 1 falling edge.
Roughly what I was thinking. What voltage do we consider, as the 5V decays, to be the limit?
I've seen 4.8 and 4.75 as the lower limit. Does your scope allow a screen capture? See if you can just capture the slope.
-
The only time I get issues with corrupt sd cards on raspberry pi is when I attempt to upgrade RRF using sudo apt-get update, sudo apt-get upgrade.
(They not actually corrupt, it just doesnt end in success)Turning power off without shutting down pi causes less problems maybe.
-
@gtj0 said in Can-o-Worms: Pi Shutdown vs Power Off:
@Danal said in Can-o-Worms: Pi Shutdown vs Power Off:
@gtj0 said in Can-o-Worms: Pi Shutdown vs Power Off:
Connect 12V to channel 1 and 5v to channel 2. Set the scope it to trigger on channel 1 falling edge.
Roughly what I was thinking. What voltage do we consider, as the 5V decays, to be the limit?
I've seen 4.8 and 4.75 as the lower limit. Does your scope allow a screen capture? See if you can just capture the slope.
Setting it up now.
-
Best I've gotten so far. Gnd to the ground to the common ground that everything shares, tip of 10x probe to pin 2 on expansion (5V). You can see the trigger setup in the upper right.
I can't go any slower on the horizontal without going into "slow acquisition mode", which I will try next.
Anything specific you were looking for?
-
@Phaedrux said in Can-o-Worms: Pi Shutdown vs Power Off:
Modern file systems and flash storage devices just aren't as susceptible to power loss corruption as older filesystems on spinning magnetic storage.
That's certainly true of SSDs, which are designed to be tolerant of power failures. I'm not at all sure that it is true of SD cards, and they don't have the possibility of storing on-board power to enable them to shut down cleanly. So there may be a real danger that if you remove power while you are writing data to them, they could get corrupted. However, a Pi that is idle following a print probably isn't writing anything to the SD card; so that risk should be low. Also, the journaling file systems used by Linux are designed to be tolerant of unexpected removal of power.
-
Exactly, that's the reason Raspbian (and thus DuetPi) use ext4, which is a mature FS. Back in the days of ext1/ext2 power failures could provoke data loss but TBH I haven't managed to corrupt a single SD card or damage a Pi 3 or 4 just by plugging the power cord. When I did my initial tests with the Duet 3 attached, I even noticed that the Pi was staying on for a short moment after the power has been cut and sent a last message to DWC, so it may be even possible that the Linux kernel recognizes this and does a final flush before the power is completely gone.
-
Ext4 uses a mechanism called "write barriers" which ensures writing metadata correctly to the file system in case of power loss. Even for volatile write caches. This prevents data corruption.