First issue after months of smooth sailing with 15.1

[donallen]

I'm running 15.1 on a 6-core AMD system I built years ago. It has 32 GB of memory and a 256 GB NVME SSD, GPT and ZFS. Also in the case is a 2 TB Seagate mechanical SATA disk, that I use of local backups and archives, also GPT and ZFS. This system has been running for 2 or 3 months flawlessly since I installed FreeBSD, replacing the Arch Linux system that was installed before.

Flawlessly until today. While I'm working, I periodically run a backup script that rsyncs my home directory with a backup directory on the Seagate. Today, when I tried to do this, the script just hung. The chatter on the console said that there was a catastrophic hardware failure on the backup disk. I tried to reboot the system but was unable to do so. Finally, I pushed the reset button and that seemed to start the reboot process but that hung with a blank screen. I tried that twice, no luck. So I power-cycled the system. It came up normally. The ZFS pool on the Seagate got mounted properly and everything is there. I was able to run my backup script and have been using the system for a few hours since the hiccup without any issues.

Because of the inability to reboot with the reset button and that power-cycling the machine was necessary to restore order, I suspect that this was a hardware issue or perhaps a driver issue getting the hardware into a wedged state. Anyone seen anything like this? This machine has been dead reliable for years (yes, I do understand that past performance is not a blah, blah, blah) and was absolutely reliable running Linux until I installed FreeBSD a few months ago.

Other than this, I've been very happy with 15.1. I've tried FreeBSD off and on for 20 years and always ran into a show-stopper. Other than some documentation issues during installation, 15.1 has really been a pleasure. And it let's me leave behind the many rough edges of Linux, e.g., the truly awful audio stuff.

/Don Allen

 

[richardtoohey2]

Anyone seen anything like this?

Not specifically like this, but this sort of thing is horrible - you don't know what to trust.

As you suspect, it seems most likely to be hardware. Check all cabling, re-seat RAM etc. and see how you go. Disk-checking tools.

I've had disk intensive tasks hang or stall with no messages but here you've actually got a message that the OS thought there was a hardware error. So either it was mistaken (bug, loose cable) or there really is a hardware error.

 

[Maturin]

(yes, I do understand that past performance is not a blah, blah, blah)

What are you excusing for, for no reason?
The specs of your machine you've given are way more than sufficient, pretty good in fact.

256 GB NVME SSD, GPT and ZFS.

Exactly what my system is installed to, and still 166 GB free. (my /home is outsourced to another drive)
You may need more RAM and maybe a faster CPU with more cores, if you do heavy load stuff like editing large high resolution video files. Unless you do this, why should you have computation power you neither need nor use?
I made a mistake when I built my new (and current) PC five years ago to gave it 64 GB RAM. Looking back it was wasted money, because I figured out for what I do with my machine, I don't need that much RAM. 16 is what I need, so 32 had been way more than enough reserve for my needs. It says, unused RAM is wasted RAM. But I guess if my kernel could talk it would say, what the heck shall I do resp. you want with so much RAM? I bet most of my RAM cells are still virgin, never seen a single bit ever.

However,
to me this seems more to presume a hardware problem of your 2TB HDD to be examined first.
Could be your disk is dying, could be simply a loose contact (check the wires!)
I guess the HDD is part fo your /etc/fstab
If any partition written in there is not detected during boot time, because the drive is missing/failing, the system will not come up.

 

[drhowarddrfine]

The chatter you saw would have been nice to see but, like most of us, we never get to grab that at the time. The output of dmesg might have shown something. And coincidences are coincidences when something crashes.

If possible, swapping out or pulling drives and ram might prove something. Especially if you have a free drive you can reinstall FreeBSD on but it's highly unlikely to be a FreeBSD since we all run it personally and professionally.

Not that it matters but my system is still running from 2004. Last year I brought up my old i386 system from my old, musty basement and was able to boot it up and search for files.

 

[donallen]

What are you excusing for, for no reason?
The specs of your machine you've given are way more than sufficient, pretty good in fact.

There was a damned good reason and it has nothing to do with the specs of the machine. My point was that the fact that the machine has been absolutely reliable for years does not guarantee that it is reliable NOW.

Exactly what my system is installed to, and still 166 GB free. (my /home is outsourced to another drive)
You may need more RAM and maybe a faster CPU with more cores, if you do heavy load stuff like editing large high resolution video files. Unless you do this, why should you have computation power you neither need nor use?
I made a mistake when I built my new (and current) PC five years ago to gave it 64 GB RAM. Looking back it was wasted money, because I figured out for what I do with my machine, I don't need that much RAM. 16 is what I need, so 32 had been way more than enough reserve for my needs. It says, unused RAM is wasted RAM. But I guess if my kernel could talk it would say, what the heck shall I do resp. you want with so much RAM? I bet most of my RAM cells are still virgin, never seen a single bit ever.

However,
to me this seems more to presume a hardware problem of your 2TB HDD to be examined first.
Could be your disk is dying, could be simply a loose contact (check the wires!)
I guess the HDD is part fo your /etc/fstab
If any partition written in there is not detected during boot time, because the drive is missing/failing, the system will not come up.

 

[donallen]

Thanks for the replies. After many years of OS development work myself, plus never running into this problem with Linux, which I ran on this hardware until a few months ago, I think the likelihood is that there is a driver bug. This happened once in three months of running FreeBSD and the cost was a few minutes to get the system running again. I do not believe any files were lost. So it's not a big problem and I will continue running FreeBSD, which I prefer, unless and until this becomes a more frequent issue, at which point I'd be forced to reinstall Linux, if only as a debugging measure (to see if it exhibits the same issue).

Again, thanks for the help.

 

[Maturin]

There was a damned good reason and it has nothing to do with the specs of the machine. My point was that the fact that the machine has been absolutely reliable for years does not guarantee that it is reliable NOW.

Maybe I misunderstood your "excuse" then. Sorry.
But anyway, it's hard to believe for me those probs you describe are caused by the system.
As I said, by what you descrobe in your OP this smells pretty as a hardware problem to me (HDD, connector, wiring).

 

[donallen]

Maybe I misunderstood your "excuse" then. Sorry.
But anyway, it's hard to believe for me those probs you describe are caused by the system.

A subtle driver bug, such as a timing issue, could do it, as I said.

As I said, by what you descrobe in your OP this smells pretty as a hardware problem to me (HDD, connector, wiring).

No question that that is also a plausible explanation. As I said in my previous post, if this becomes a serious issue for me, I will have to switch back to Linux. If Linux exhibits the same issue, then either it's hardware or FreeBSD and Linux have the same driver bug. If it does not exhibit the issue, then the driver bug becomes the likely culprit.

 

[CShell]

So I power-cycled the system. It came up normally. The ZFS pool on the Seagate got mounted properly and everything is there.

Not stopping you from going back to Linux if that is what you want to do...

But what might have happened is that your machine picked up a few (pkg?) updates and needed a reboot. It reads like 'you rebooted' and all is well ! That tells me a power cycle was good !

I run Linux and FreeBSD side by side (literally) and I can tell you that Linux is FAR from a slam dunk when it comes to updating, upgrading, etc. I have far less problems on my FreeBSD systems because I can set ZFS snapshots and other things before FreeBSD updates / upgrades.

One difference is that I use UFS filesystems for my backups/rsync and I have zero problems when I do that. I am NOT saying that you should not use ZFS for backups, and ZFS does seems to work for you, but just reporting that I have excellent experience using UFS.

 

[fjdlr]

Using send/receive on zfs system
See syncoid