how to prevent boot volume to corrupt

[Keven]

Hi,

I have that notification that appear on January 7th:

The boot volume state is ONLINE: One or more devices has experienced an error resulting in data corruption. Applications may be affected.

actually my freenas system did not seem to have any trouble to boot, but seeing that notification doesn't feel right doing nothing about it. So, I ordered another thumb drive: Sandisk Ultra Flair 16GB to make my boot pool mirror with my sony 8GB. At first i thought that it would enable the OS to not only see the corruption of one drive but correct it ( a little bit like it does with my data pool with RAIDZ) but the more i read about that the lesser i think that it will do that. Seems like it is only useful when a drive fail completely then it will make the switch and boot from the other one.

So, I am wrong? what is the best solution to prevent/repair corruption as it goes with the years...


I tried to slam some ECC modules on my thumb drive to make it ECC , but did not work so well...

 

[Chris Moore]

actually my freenas system did not seem to have any trouble to boot, but seeing that notification doesn't feel right doing nothing about it. So, I ordered another thumb drive: Sandisk Ultra Flair 16GB to make my boot pool mirror with my sony 8GB. At first i thought that it would enable the OS to not only see the corruption of one drive but correct it ( a little bit like it does with my data pool with RAIDZ) but the more i read about that the lesser i think that it will do that. Seems like it is only useful when a drive fail completely then it will make the switch and boot from the other one.

It can't correct from a data error unless there was a redundant disk at the time the error happened. Adding a disk after the error does nothing for you because there was no redundant data to correct the error from. USB 3.0 memory sticks are quick, and quick to fail. If you just use USB, you should use USB 2.0, but SSDs are the recommendation.

Hardware Requirements
http://www.freenas.org/hardware-requirements/

 

[Keven]

Sorry for not being clear I know that it will not correct the error after the fact I plan save again the config file and make a clean install for both USB stick.

My question was if putting 2 USB stick mirror for boot volume would prevent such non correctable error from happening in the future?

And for the ssd... Not a option because of the uses of sata port I plan to grow my pool in the future and I will need those precious sata port for data drive

 

[Chris Moore]

My question was if putting 2 USB stick mirror for boot volume would prevent such non correctable error from happening in the future?

Often times, failures in USB memory modules are catastrophic in such a way that there is no way to repair or recover. They are just not reliable. Every USB memory module I have ever used for boot media has failed within a year. That is why I just don't use them any more.
The smallest capacity SSD from a quality company like Intel is going to be much better and last for a decade.
If you don't have room in your build for a SSD, you could use something like this:
https://www.newegg.com/Product/Prod...dom&cm_re=sata_dom-_-9SIA5EM54X6826-_-Product

 

[Keven]

The smallest capacity SSD from a quality company like Intel is going to be much better

I understand that ssd are much more reliable, but like running a zpool in stripe it's not able to repair corruption in the unlikely event that it happen. So, there is no way to have redundant boot pool in the way that it can handle error and correct it like the data pool

if you don't have room in your build for a SSD, you could use something like this:
https://www.newegg.com/Product/Product.aspx?Item=9SIA5EM54X6826&Description=sata dom&cm_re=sata_dom-_-9SIA5EM54X6826-_-Product

It's Not about room, but about the limited number of sata port available on the motherboard. If I can get away without buying an HBA adapter it would be great.

 

[Chris Moore]

So, there is no way to have redundant boot pool in the way that it can handle error and correct it like the data pool

Sure. You can have redundant USB boot media, and many people do, but I have found them to be less reliable. It may run for a long time without fail, but when it does fail, you will likely need to install new media and reinstall a fresh copy of FreeeNAS and restore your configuration database from a backup. USB media is just not robust like disk media. It doesn't fail in predictable ways and it provides no predictive information through SMART tests like normal drives. Use it if you must. It is an option that many people choose.

 

[Basil Hendroff]

It is an option that many people choose.

I use mirrored USB boot media exclusively on my FreeNAS servers, but I have to agree with @Chris Moore that they are prone to unexpected and inexplicable transient behaviour. Probably the one hardware component on my servers that I consciously have to keep an eye on.

 

[Redcoat]

Seems like it is only useful when a drive fail completely then it will make the switch and boot from the other one.

@Chris Moore, I smell OP confusion about "redundancy" and "failover" here (and it's frequent in the forum, I think, but not so often addressed).

Back in the day when I was comissioning my backup Dell and consuming USB sticks fast I expected failover but did not get it, learning the distinction with redundancy in the process. Each time one of a pair of mirrored USB boot drives failed I had to identify the failed drive and pull it in order to boot from the other (when I could again mirror a further fresh USB stick). I have not used USB for a couple of years now so I can't speak to the precise behaviour today.

 

[Chris Moore]

I have not used USB for a couple of years now so I can't speak to the precise behaviour today.

I have not used USB for booting since prior to 2016, but I don't recall exactly when I made the switch. I have been using my current boot hard disk drive mirrored pair since 2016.

When I did use USB last, it didn't appear to make a difference which USB drive failed... When one failed, the server would crash. I didn't need that to happen more than a couple times before I was ready to give something else a chance.
I have had zero problems with spinning disks as boot drives. I use a mirrored pair of 2.5" drives that were made for a laptop. I imagine SSDs would be even more reliable.

 

[pro lamer]

doesn't feel right doing nothing about it

I'd check what files were affected. just for sanity...

I'm trying to imagine what might cause USB drive corruption of type the OP had. Maybe overheating?

Anyway mirrored pendrives could help such cases, providing the system boots (see answers above)

Sent from my phone