Moving external drives into TR-004, but TrueNAS not seeing them.

[SaltyCoffee]

Just as an update to this.

I solved it by configuring the RAID in QERM. I was then able to configure a 1 disk pool in TN that can access the full space of the RAID I configured in QERM. Once I feel confident about this, I'll move all the data from the original pool to the new pool and be done with this episode. I don't mind outsourcing the RAID at this point. I'm happy to continue using TN as a plugin manager. And I still get to use ZFS.

 

[sretalla]

I don't mind outsourcing the RAID at this point. I'm happy to continue using TN as a plugin manager. And I still get to use ZFS.

Maybe I missed somewhere where you said you don't like keeping your data, but what you're proposing is a road to data loss.

You will possibly be able to see corruption as it happens to your pool (thanks to ZFS checksumming), but you will be able to do nothing to correct it.

Read this carefully before you put any data in that pool:

What's all the noise about HBA's, and why can't I use a RAID controller?

This is relevant to FreeNAS and TrueNAS CORE. Some parts of it might also be relevant to Scale, but I don't really know how reliable the Linux drivers are. 1) An HBA is a Host Bus Adapter. This is a controller that allows SAS and SATA devices to be attached to, and communicate directly with...

www.truenas.com

 

[SaltyCoffee]

Maybe I missed somewhere where you said you don't like keeping your data, but what you're proposing is a road to data loss.

You will possibly be able to see corruption as it happens to your pool (thanks to ZFS checksumming), but you will be able to do nothing to correct it.

Read this carefully before you put any data in that pool:

What's all the noise about HBA's, and why can't I use a RAID controller?

This is relevant to FreeNAS and TrueNAS CORE. Some parts of it might also be relevant to Scale, but I don't really know how reliable the Linux drivers are. 1) An HBA is a Host Bus Adapter. This is a controller that allows SAS and SATA devices to be attached to, and communicate directly with...

www.truenas.com

Interesting. Thanks for the link. So is my take-away correct in that, the issue here is ZFS working with a raid card that isn't designed to handle the overhead generated by ZFS? How would one go about testing the actual performance?

 

[jgreco]

I solved it by configuring the RAID in QERM. I was then able to configure a 1 disk pool in TN that can access the full space of the RAID I configured in QERM. Once I feel confident about this, I'll move all the data from the original pool to the new pool and be done with this episode. I don't mind outsourcing the RAID at this point. I'm happy to continue using TN as a plugin manager. And I still get to use ZFS.

Yes, you still get to use ZFS, until something happens and ZFS can't fix a problem because you have no redundancy at the ZFS level.

 

[jgreco]

is my take-away correct in that, the issue here is ZFS working with a raid card that isn't designed to handle the overhead generated by ZFS?

No. That's just one of a number of issues.

 

[HarryMuscle]

I only skimmed through all the posts so my apologies if you already answered this.

Is the issue you're describing only affecting the GUI? In other words, when using JBOD do both drives show up via the command line as attached devices (ie: /dev/sda, /dev/sdb, or similar) but the GUI only displays one of them? Or even via the command line is there only one device in total?

Thanks,
Harry

 

[SaltyCoffee]

Soooo.... I thought I'd update this thread, 6 months later. The ongoing saga that is TN parenthood.

I'd like to credit the helpful souls here that the USB set up is definitely not a long term solution; CONFIRMED. You can visibly see the damage this set up does to the drives over time. And in the spirit of full disclosure, this thing is running < recommended RAM (6GB), and non-ECC on top of that. Also running at 93% capacity haha (did I get all the newbie hits there?). It has utterly ravaged these external USB drives to pieces. It's quite impressive. And I'm the only one using this system for maybe a couple hours of PLEX time a day (and the traffic that comes with aggressively updating the library with content on a regular basis). The drives are still working, but I have zero confidence in them making it to the end of the year. The data itself seems to have miraculously maintained integrity, though (no idea how). Not one lost file to corruption yet... (yet...).

As even I'm concerned with this set up now, I've moved the drives (boot and pool) to an old Mac Pro. Which I was reading constantly here that it doesn't work, and there are plenty of people who found that to be true. I was quite lucky, in light of that, that this was a plug and play solution for me. Dropped the drives in and booted it right up. Just had a vnet issue with the plex jail, which was easily fixed by just changing the vnet interface to the ethernet and then back to auto. Plex came right up after that. Drives are resilvering as we speak. And getting the drives back on a proper SATA bus even brought one back that was faulting constantly. And another that was degraded, came back to online status. So the drives collectively gasping a sigh of relief. Happy to be running ECC RAM and CPU now too (xeon with the minimum 8GB mem but more is coming). I've ticked all the best practices boxes with the move (well, no IPMI). The only issue is of ventilation.This brings me to my question, at this thread...

Moral of the story is, if you absolutely, positively got to run TN w external USB drives, keep usage below 50%. After about 60% utilization is when the walls started to burn. By 80%, errors were constant. And by 90% drives started failing.

And, that's that.

Bonus question, irrelevant but: If anyone can suggest a good (meaning any) Mini-ITX mainboard that checks all the boxes for TN and has a LGA1366 socket, please let me know! Seems to be a bit of a unicorn that one....

 

[HoneyBadger]

Mini-ITX mainboard that checks all the boxes for TN and has a LGA1366 socket

That's a fairly old socket at this point, and the very limited marketspace of that crossover (ITX + X58 chipset) seems like you could probably get an LGA1356 or LGA2011 system for less than what it would cost to buy the motherboard alone, and reap the benefits of faster processing/better general platform support.

Adding onto that, I couldn't find an ITX system - the closest I found was the Shuttle SX58H7 barebones machine.

Depending on the age of that Mac Pro (model number ranges from 2009-2013) you might even be able to sell it to a retro Apple enthusiast for more than enough to buy a more modern server solution. :)

 

[SaltyCoffee]

That's a fairly old socket at this point, and the very limited marketspace of that crossover (ITX + X58 chipset) seems like you could probably get an LGA1356 or LGA2011 system for less than what it would cost to buy the motherboard alone, and reap the benefits of faster processing/better general platform support.

Adding onto that, I couldn't find an ITX system - the closest I found was the Shuttle SX58H7 barebones machine.

Depending on the age of that Mac Pro (model number ranges from 2009-2013) you might even be able to sell it to a retro Apple enthusiast for more than enough to buy a more modern server solution. :)

I'm looking at the Atom based systems (Denverton), currently. Those seem to check all the boxes, and then some. Just not sure about the CPU clock speeds on them (max out at 2.4Ghz when plex wants at least 3Ghz) or how well they handle transcoding. Need a bit more research. But, I was looking for a 1366 board because I happen to have a couple of these old Xeon processors (W3540) in hand and they're enough for what I need. Seemed a shame to buy more cpus when I have those. But I read somewhere that the ITX boards at the time these cpus were relevant couldn't handle the power draw they generated (I think it's like 135w or something outrageous like that). So I imagine I won't find any.

This Shuttle system just nearly hit the mark though, well done. Only real issue I see is the Realtek NIC. And maybe it wouldn't be spoiled by a few more SATA connectors. I wonder if these i7 1366 sockets would be compatible with an old Xeon? If it weren't for that nic though, nice find.

Re: selling the mac pros, it never even crossed my mind. You think that's feasible? I have 2 of them. Worth looking into at the least, thanks for the idea.