[SOLVE]UNRAID HANGING WHEN STARTING ARRAY ONLINE and START PARITY-SYNC

[fazrul]

Hello everyone,

I been stuck for several days and been checking the net to find solution and seems the issue still persist.

 

How to reproduce:
using SATA cable 

add one parity with 1 disk 2TB (same size) and 512GB SSD - no go (hang at 2%)
add one parity with 3 disk 512GB SSD - no go (hang at 1%)

 

Internal Mini SAS 36p SFF-8087 M Latch to 4X SATA 7p F Breakout Cable

one parity with 1 disk 2TB and hang at 10% during parity sync
One Parity with SSD connected and 3 SATA connected hang around 5% during parity sync
One parity with 3 SATA 2TB, hang at 7$ parity sync

Expected results: parity Sync completed

Actual results: parity Sync hang 

 

Other information:

perform memtest for 6 hours and not seeing any issue
notice while using Mini SAS to SATA the parity sync speed drop from 180MB+ to 14MB before freeze 
using safe mode the parity sync will take approx 1 days plus with 17MB speed.

feedback_diagnostics_1655979140.zip syslog

Edited June 24, 2022 by fazrul
issue resolved

[trurl]

10 minutes ago, fazrul said:

perform memtest for 6 hours and not seeing any issue

Your syslog certainly has a lot of entries about RAM.

 

Builtin memtest can't check ECC memory. You will have to get the latest official memtest86

[itimpi]

As was mentioned millions of RAM related errors in syslog.    I would not be happy until the cause of those is tracked down and eliminated.

[trurl]

7 hours ago, itimpi said:

not be happy until the cause of those is tracked down and eliminated.

You don't even want to attempt to run a computer unless RAM is reliable. 

[JonathanM]

10 hours ago, fazrul said:

add one parity with 1 disk 2TB (same size) and 512GB SSD - no go (hang at 2%)
add one parity with 3 disk 512GB SSD - no go (hang at 1%)

SSD's in the parity array isn't recommended for performance reasons, unless all the array disks are SSD including parity, even then there are limitations.

[fazrul]

11 hours ago, trurl said:

Your syslog certainly has a lot of entries about RAM.

 

Builtin memtest can't check ECC memory. You will have to get the latest official memtest86

Thanks for clarification. I've notice the error about RAM but a doubt that could be the issue since it was running fine under Truenas and previous unraid setup.

I've remove all ECC RAM and just use 16GB for now. After 5 hours later the result PASSED. Hence I will only use one RAM to perform parity sync and awaiting next 2 hours to complete and will update the outcome.

 

 

 

3 hours ago, trurl said:

You don't even want to attempt to run a computer unless RAM is reliable. 

Yep got it thanks for that information. Did use this setup few years back with Unraid but didnt encouter any issue. It's seems reusing the setup with new config shows there an issue with RAM.

 

2 hours ago, JonathanM said:

SSD's in the parity array isn't recommended for performance reasons, unless all the array disks are SSD including parity, even then there are limitations.

Thanks, there was a typo as what i meant was using cache as SSD.

 

Anyway will update the result 

[fazrul]

Seems doesn't solve the issue even the memory result has passed.

 

 

I saw this message and change the date and time on BIOS but it still hang

 

Line 3913: Jun 24 02:48:30 Tower ntpd[1561]: kernel reports TIME_ERROR: 0x41: Clock Unsynchronized
    Line 3929: Jun 24 02:50:04 Tower nginx: 2022/06/24 02:50:04 [error] 1895#1895: nchan: A message from the past has just been published. Unless the system time has been adjusted, this should never happen.

syslog MemTest86-Report-20220623-150250.html

[fazrul]

i fixed the issue it is due to bad RAM slot evendo the memtest that i did on slot 12 are working fine.

 

Thanks everyone for the assistance.