Crashing with 6.92

[ez009]

I've seen several similar threads but thought i should post my info in a separate topic (?)

Since updating to 6.92 my server won't stay up for more than 8 hours at a time.
It's not shutting down per se but i lose connectivity and have to power it down to reboot.
After reading other threads I enabled syslog mirroring to flash so i have the latest syslog and diagnostics that should have some details.
(it crashed around 4am)

I think I might try reverting to the previous stable version unless that is unadvisable for some reason? Pls let me know.

thx


 

syslog tower-diagnostics-20210421-1830.zip

Edited April 22, 2021 by ez009

[ez009]

update:
after powering up after the last crash to get the recent syslog info - i attempted a clean reboot but now the server won't power up.
i get a momentary flash from the power lights, the power supply's are running (fans on etc) 

So maybe it's a co-incidence that i recently upgraded to 6.92? - and i have a hardware issue.
only things i saw in the log were lots of
"Tower kernel: w83795 0-002f: Failed to read from register 0x03c, err -6"

and "kernel reports TIME_ERROR: 0x41: Clock Unsynchronized"

I've tried clearing cmos and replacing cmos battery, pulled all ram except for one stick - with same result.
this is a Supermicro - X8DAH in a 24-bay with redundant 720pws.

This system has been up & running stable for over 3 yrs with a couple upgrades to processors and ram about a year ago.
If anyone has any troubleshooting advice or can tell anything from the logs let me know.  - thx

[JorgeB]

9 hours ago, ez009 said:

and i have a hardware issue.

First fix that then see if it solves the problem, don't see any reasons for the crashing logged.

[6of6]

Having been through it so many times... PLAN... UPGRADE... (BACKUP?)...  It's the wrong order, isn't it?

 

It seems so simple. The only reason I haven't restarted my server in the last 7 days is that I don't trust it to restart.

 

What does that say?!

 

6.

[ez009]

So I stripped it down to the motherboard only, out of the case with a new power supply, new cmos, swapping cpus and ram in a minimal config.
It appears something in my mb is toast as i can not even get it to post.
Spent 48 hours trying to troubleshoot but have conceded, and a replacement mb is on the way along with 2 new cpu coolers (sm active).
I think perhaps my upgrading to 2 - x5860's while using the stock passive heatsinks was probably putting some additional heat/stress on the mb.

and these components are like 10 yrs old..   anyway, I'll check back next week to confirm that everything's fine.

thx

[ez009]

the saga continues.....

I eventually got the new motherboard.. up and running. It'd been years since I'd been in the bios but i think at this point I have the settings correct.
(I am dealing with some RAM not being recognized so (for now) I am only using CPU2 which recognizes all DIMMS on that side)

Unfortunately I am back where I left off with the 6.9.2 freezing.  In fact it was freezing within 15mins of logging into the GUI.
At first I wasn't sure if it had to do with the hardware, bios, or several configuration changes.

the syslog would sometimes show lots of these before freezing:
May 24 01:05:39 Tower kernel: w83795 0-002f: Failed to read from register 0x03c, err -6
May 24 01:05:39 Tower kernel: w83795 0-002f: Failed to read from register 0x034, err -6

*note that the freeze won't happen until I actually log into the GUI

I only started the Array twice(?) during testing and have since opted to make sure unraid will stay alive on its own before I move any further.
Speaking of moving further, I realized that my prior stable installed build was in fact 6.8.3(!) -  so I made a backup of the current 6.9.2 flash drive

and downgraded to 6.8.3...

But Same issue - freezing within the amount of time it takes to peruse the previous syslog. I am also seeing similar errors.

*One note of concern, my cache pool is no longer recognized after the downgrade.. all 4 of those SSDs are now unassigned

 

Next, I created a fresh Trial install of 6.8.3 on a new USB drive. I disconnected my SAS card and booted up.
After logging in, it's been up and running for over 3 hours now. 

I'm wondering how to proceed? I was planning on copying super.dat to the fresh install, install the SAS card and test the Array,
but with my cache pool not being recognized after the 6.8.3 downgrade, I'm concerned.  (it was recognized fine in 6.9.2)
If i copy super.dat from the 6.9.2 flash into this fresh 6.8.3 install, would that work??? - or should i try i clean install of 6.9.2?

 

The Plex docker is the main thing of importance I'd like to preserve in addition to the array.

 

Edited May 26, 2021 by ez009

[JorgeB]

2 minutes ago, ez009 said:

but with my cache pool not being recognized after the 6.8.3 downgrade, I'm concerned. 

You just need to re-assign all the previous pool members, order is not important, also there can't be a "all data on this device will be deleted at array start" or similar warning next to any of the pool devices.

[ez009]

ok, i started to try that but it seemed i was only able to add one drive.  so i unassigned it again. Wasn't sure if it would be reformatted
Also the old config was crashing so fast i didn't have much of a chance to do anything.
So should I copy the super.dat though? and would it be ok to copy the one from my 6.9.2 config into this fresh 6.8.3?

[JorgeB]

You can copy super.dat but that just contains the array assignments, cache still needs to be re-assigned.

[ez009]

I see the cache slots now. should be fine. However, i guess copying the super.dat doesn't work for this Trial version i'm playing with..
so I guess i will need to copy the new install to my old usb drive. or I may buy a 2nd license..  
In either case, is there anything else that would be considered "safe" to copy over from my old Flash directories in order to preserve as many settings as possible?

for example (in config): shares, pools, disk assignment, various other .cfg files?

[JorgeB]

7 minutes ago, ez009 said:

i guess copying the super.dat doesn't work for this Trial version i'm playing with..

Yes, I forgot that, you can't use an existing config with a trial key.