Finding errors after replacing some drives

[JorgeB]

1 minute ago, icedragonslair said:

This way I can kind of look through it myself when anything pops up.

If you have notifications enable you'll get one, but read errors are logged like this:

 

Quote

Aug  4 14:52:55 mediaserver1 kernel: md: disk18 read error, sector=183334984

 

And jus below there's this:

 

Quote

Aug  4 14:52:55 mediaserver1 kernel: md: recovery thread: multiple disk errors, sector=183334984

 

Which in unRAID speak means:

"there are errors in more disks than current redundancy can correct, the rebuild/sync will continue but there will be some (or a lot) of corruption."

 

[icedragonslair]

1 hour ago, pwm said:

 

You could use the Dynamix File Integrity plugin.

After some reading on the subject (couldn't find any current info newer than 2016), I am assuming that this is still supported.

 

I also didn't see if that would check the file integrity to see which files are corrupt at this time, or is it a run and from then on situation?

[pwm]

8 minutes ago, icedragonslair said:

After some reading on the subject (couldn't find any current info newer than 2016), I am assuming that this is still supported.

 

I also didn't see if that would check the file integrity to see which files are corrupt at this time, or is it a run and from then on situation?

 

The checksums must be have been created before the suspected introduction of corruption. Then a new checksum can be computed and compared with the original value to figure out if any changes have been introduced to the files.

[icedragonslair]

I think I have enough info to tackle this once the new drive comes in

 

Plan of attack 

 

I am going to mount the old drives (it's actually drives 7 & 13 that I still have) one at a time, do a checksum on each drive, compare those and replace the files that need to be replaced, or should I just do the full copy? 

Then for drive 5 I will just try to replace everything that's on it, if I can.

Then Replace drive 18 and rebuild the parity.

 

Does this sound right?

 

Again thanks so much,

 

Ice

[JorgeB]

11 hours ago, icedragonslair said:

replace the files that need to be replaced, or should I just do the full copy? 

Either way works, full copy would be faster but if you're plan to create checksums for the future might as well do it now and replace only the corrupt files.

[icedragonslair]

LOL, well the board died after the first disc transfer, (still under warranty) go figure.

I am correct in assuming that if i replace it with another board and cpu, and as long as they are supported by unraid that it wont matter to the OS, correct?

If that's the case I'm just going to buy a new one and use the warranty repaired one as a shelf spare.

If it I'm wrong chime in and let me know 

 

Thanks loads for all,

 

Ice

 

 

[pwm]

10 minutes ago, icedragonslair said:

I am correct in assuming that if i replace it with another board and cpu, and as long as they are supported by unraid that it wont matter to the OS, correct?

 

Yes, as longs as the board/cpu is supported, it's ok to move the unRAID installation between different machines. If you do hardware pass-through to VM, then that represents additional compatibility requirements.

[icedragonslair]

1 hour ago, pwm said:

 

Yes, as longs as the board/cpu is supported, it's ok to move the unRAID installation between different machines. If you do hardware pass-through to VM, then that represents additional compatibility requirements.

 Nah, I just do a simple straight-forward setup, no vm's at all. Basically just a home media storage server, basically a giant nas, for my movie collection. I have a second pro license , so someday i will play around with the other stuff, but not for the foreseeable future

 

This one was a z97 intel 1150 chipset with an i5-4460, 16gb ddr3 and 1kw power supply (ample lol). so i am just looking to expand the case and get a spare board (may have to go with a newer model board since the cost of these older boards have skyrocketed. Would love to go with a server board that has 3 x pci-e 8x slots, but those too are very pricey.

If I upgrade i just need to make sure there is driver support in linux, correct (I won't be using current generation)?

For now it looks like ill be running the same thing as before. 

 

Thanks for all,

Ice

[icedragonslair]

Sorry to necro my old thread, but i just got my repaired board in yesterday (gigabyte rma - bios issue...sigh) and I decided to remove everything, clean thoroughly and re-assemble the server. And I now have two more; 

1. I am assuming that I can re-arrange the cables to the drives to neaten it up and that unraid will pick them up via serial #, correct?

2. Am I better off keeping the cash on the mobo sata connectors instead of on the raid cards?

I am finally getting this back together and hopefully running again...missing my media  :)

Keep your fingers crossed for me  :)

Thanks for all,

Ice

 

[JorgeB]

2 hours ago, icedragonslair said:

1. I am assuming that I can re-arrange the cables to the drives to neaten it up and that unraid will pick them up via serial #, correct?

Yes

 

2 hours ago, icedragonslair said:

2. Am I better off keeping the cash on the mobo sata connectors instead of on the raid cards?

Depends on the controller but usually yes, mobo ports are better since they always support trim.

[icedragonslair]

 

Quote

 

2. Am I better off keeping the cash on the mobo sata connectors instead of on the raid cards?

Depends on the controller but usually yes, mobo ports are better since they always support trim.

 

 

 

So the plan will be to keep the cache and parity on the mobo controller and all the data drives on the cards...:) (with added 2 drives filling in 2 of the slots on the mobo).

Thanks again for all, will touch base when up and running 

 

Ice

[icedragonslair]

I am about ready to give up and scrap the server. It seems i am cursed!!!

First I get my board back and install, I replace the data on the other drives, but before I can replace the one drive with pending sectors (Disk 18, that started all this crap), another (Disk 15) pops up with pending sectors and read errors. So i decide to replace that one since I have a copy of the data that is on it, and once that is done and data reloaded, it throws another (Disk with pending sectors and read errors. It doesn't matter whether they are new or old drives, is it just a cascade thing following the errors around the drives. I am at wits end and ready to just dump the server altogether, give up one this crap and sell my hardware & keys.

HELP
 

mediaserver1-syslog-20180914-2335.zip

[JorgeB]

A faulty power supply can in some cases damage disks, including causing bad sectors, so if you have a spare swap it, other than that or some other obvious issue like extreme vibrations it could just be bad luck.

[icedragonslair]

I stress tested the power supply and it is  well within specs, but I do not have an o-scope to check for spikes and I don't have a second one lying around (if another one goes I'll rma it).  Otherwise I'll see if I can keep doing this...fixed income and a grand spent out of budget this past month probably means it will be a flower pot for the next 6 months

 

Thanks for the quick response,

 

Ice

[JorgeB]

Looking at the earlier diagnostics you posted, disk15 looked fine, but disk8 was showing some warnings, as is disk12, doesn't mean it's about to fail but keep an eye on it, especially the Raw_Read_Error_Rate attribute and any more UNC errors.

[icedragonslair]

Thanks,

If I strip the data from disk 18, 8, & 12 and exclude them from the shares. Then run a parity check and if this shows only problems with these disks, replace them and do a rebuild. then include them in the shares, reload the data and rebuild parity...I should be fine???

 

Will this work?

 

[JorgeB]

You could rebuild them one by one, on the first rebuilds there would likely be read errors on other disks but since they would be empty not a big deal, but IMO if you need to replace more than 1 disk and they have no data might as well do a new config, replace them all at the same and rebuild parity.

[icedragonslair]

26 minutes ago, johnnie.black said:

You could rebuild them one by one, on the first rebuilds there would likely be read errors on other disks but since they would be empty not a big deal, but IMO if you need to replace more than 1 disk and they have no data might as well do a new config, replace them all at the same and rebuild parity.

So I would first strip them and remove from shares, remove those three drives, then do a new config...(will I lose anything doing this?) and then just add the new drives as normal.

This is what you're saying?

 

thanks

[JorgeB]

You don't need to remove them from the shares, just move all the data you can to other disks and do a new config, data on the remaining data disks will be untouched, just carefull how you copy, always disk to disk, never share to disk or disk to share.

 

 

[icedragonslair]

39 minutes ago, johnnie.black said:

You don't need to remove them from the shares, just move all the data you can to other disks and do a new config, data on the remaining data disks will be untouched, just carefull how you copy, always disk to disk, never share to disk or disk to share.

 

 

Understood, and thanks loads

[icedragonslair]

On 9/15/2018 at 11:51 AM, johnnie.black said:

You don't need to remove them from the shares, just move all the data you can to other disks and do a new config, data on the remaining data disks will be untouched, just carefull how you copy, always disk to disk, never share to disk or disk to share.

 

 

I am now almost done offloading the data to the other drives (Krusader disk to disk) and so i don't mess this up I am to follow the steps below:

 

 - Once the affected disks are empty, I am just to do a 'New Config'.
 - Replace the affected disks per the norm*.

 *(or should I just replace them first and then do a 'New Config'.)

 

Thanks again for the help

 

Ice

 

 

[JorgeB]

-After copy is done shutdown and replace the old disks with the new ones.

-Power on, array won't start due to the missing disks: Tools -> New Config -> Retain current configuration: All -> Apply

-Assign the missing disks, which will be all the new ones.

-Start array to begin parity sync.

[icedragonslair]

I am assuming to preclear those disks prior to this.

 

Thanks again for all the help,

 

Ice

[JorgeB]

1 minute ago, icedragonslair said:

I am assuming to preclear those disks prior to this.

It's not required, but if the disks are new it's good to test them before adding to the array, with preclear or any other testing util.

[icedragonslair]

12 minutes ago, johnnie.black said:

It's not required, but if the disks are new it's good to test them before adding to the array, with preclear or any other testing util.

Awesome, yeah I'll preclear. Getting a little ahead of myself with this next question...it is okay if after all this is done that I reset the drives so they are in their corresponding slots in the trays, correct? (ex: Disk 13 is now in slot 9, i should be fine to swap these to their proper slot, correct?)