Help! All shares disappear, Array still online

[tillo]

Hi, twice in about two weeks now all shares have disappeared from the network, when login into unraid web interface I can verify that the Array is online, every disk seams to be okey. But when I go to the Shares tab, its blank, nothing is there. Like WTF!? where did all my shares go?

After a reboot of the unraid server everything is back to normal. Pls help.

[JorgeB]

Grab and post the diagnostics when it happens again and before rebooting.

[tillo]

1 hour ago, johnnie.black said:

Grab and post the diagnostics when it happens again and before rebooting.

There you go Johnnie

[JorgeB]

OOM errors, see if this helps:

 

 

[tillo]

Thanx Jonnie. I will read it when I get home :-)

[tillo]

I will try to implement the memory tweek recommended in the linked thread and see if that changes anything.

 

[tillo]

Hi, i did the memory tweek and I am now running 6.4.1 and I think that i still have the same problem.

@johnnie.black could you take a look at the sys log file, for me. (See next post for log file)

[JorgeB]

Try upgrading to latest v6.5rc, possibly related with this:

 

 

[tillo]

Would that be wise?

Also, I do not have a single VM on the computer, just a plain unraid with just a handful of plugins. And it looks to me that the linked thread is more related to ppl running unraid in combination with VM?

 

I would like to keep my system as stable as possible.

[JorgeB]

SHFS was the process killed, so it could be related, you don't have VMs but only have 4GB of RAM.

 

Any stability issues with the new rcs are more likely related to docker/VMs, you should have no issues running for a basic NAS server.

[tillo]

Okey, I will give it a try.

I will update this thread when I have tested it out. Thanx

[tillo]

@johnnie.black It happened yet another time 

this time the server was running with 6.5.0-rc5. See the log file below.

[JorgeB]

Again out of memory errors, this time killing the file integrity plugin, also some checksum errors there, don't know if they are expected, maybe 4GB is just not enough for the plugins you're running, v6.4/6.5 uses a little more RAM then before, try adding 4GB more.

 

 

[tillo]

okey. I will try adding a bit more RAM to it. ( it worked flawlessly with 4gb in unraid 6.0 all the way to 6.3.5 after that the problem started showing up).

Yes you are correct, I have some checksum errors. Some are expected, the ones that is shown in the log file is expected. But I had a few, about a month back that was not expected. I have been wondering about that. Would unraid correct them automatically or? What should I do in order to prevent data loss?

[JorgeB]

27 minutes ago, tillo said:

Would unraid correct them automatically or?

No.

 

Start by running memtest, it's the most common cause of data corruption for non ECC servers.

 

 

[tillo]

Just wanna give you a big thanx for your support Johnnie in all of this. It is super appreciated!

My unraid server is running on an intel server motherboard, with ECC memory. I do not wanna run any server without ECC memory, right  

 

A memtest you say. Is it a way to run a memtest directly from the unraid webGUI. Or should i download a

generic memtest and create a bootable USB stick and run it of from that?

 

[JorgeB]

I was going to check your board model before posting to run memtest, but you had already removed the diags, I thought I remembered being non ECC, but since I see so many diags every day it's normal I got mixed up , if you're running ECC no point in running memtest, you can still check the system event log for any issues but an uncorrectable memory error would (or should) halt the server preventing any corruption.

[tillo]

So, what should the next logical step be then, it it is not the memory?

[JorgeB]

Are you sure nothing is modifying the files, it could be just metadata, it's normal for media indexers to change that.

[tillo]

23 minutes ago, johnnie.black said:

Are you sure nothing is modifying the files, it could be just metadata, it's normal for media indexers to change that.

 

I think the errors that you noticed in the log file was most likely due to changes in the metadata.
But I also had a few errors a month ago, and that was on some files that should not have been modified/altered in any way. Yet it showed up as corrupt files when doing a check sum test on them. (I don't know if this is related or not to that, but the parity check sometimes tells me that I have a few mismatches in the parity data as well. Normal?)

 

Quote

Again out of memory errors, this time killing the file integrity plugin, also some checksum errors there, don't know if they are expected, maybe 4GB is just not enough for the plugins you're running, v6.4/6.5 uses a little more RAM then before, try adding 4GB more.

 

I have currently 1 unit of ktd-pe313e/4gb (kingston ECC reg) installed in the server. Just to make sure, is it anything in BIOS that I need to enable in order for the motherboard to properly use the ECC memory capability?

I also found 3 spare memory of (ktd-pe313e/4gb) that I had bought just in case I would need it in the future for the server. So I will add all of that later this week.



 

[JorgeB]

3 hours ago, tillo said:

(I don't know if this is related or not to that, but the parity check sometimes tells me that I have a few mismatches in the parity data as well. Normal?)

No, not normal, the only acceptable number of sync errors after a parity check is 0.

 

3 hours ago, tillo said:

I have currently 1 unit of ktd-pe313e/4gb (kingston ECC reg) installed in the server. Just to make sure, is it anything in BIOS that I need to enable in order for the motherboard to properly use the ECC memory capability?

Not usually, remind me the board you have.

[tillo]

Quote

  3 hours ago, tillo said:

(I don't know if this is related or not to that, but the parity check sometimes tells me that I have a few mismatches in the parity data as well. Normal?)

No, not normal, the only acceptable number of sync errors after a parity check is 0.

Humm, I will create a new post tmr, to adress this then.


 

Quote

3 hours ago, tillo said:

I have currently 1 unit of ktd-pe313e/4gb (kingston ECC reg) installed in the server. Just to make sure, is it anything in BIOS that I need to enable in order for the motherboard to properly use the ECC memory capability?

Not usually, remind me the board you have.

intel s1200v3rps

[JorgeB]

8 minutes ago, tillo said:

intel s1200v3rps

Not familiar with that model but I believe it is ECC only, so should be enable by default, you could try checking the system event log, any detected hardware issues should be logged there, if nothing there a controller or a disk are possibilities, you'll need to check if errors (checksum errors since with parity errors there's no way to know which disk they came from) are always on the same disk, on the disks on the same controller, etc to start ruling things out.

[tillo]

I started a new thread regarding the Parity errors and checksum errors.

I will keep this thread updated with more information about the problems with the of memory errors,  after that I have installed additional modules of ram into the server.