Auto Trace mode?

[Rand]

I had a small problem with my filesystem which ran full b/c freenas chose the one with an iscsi vol on top of it as system vol...

After experiencing weird issues (and some error messages) i removed the volume and rebooted... only to find i could'nt run smb (no samba config file (?) ) so i rebooted again...
After that the system rebooted itself (? maybe kernel panic) and turned on a trace mode which is currently running and blocking everything... no gui, no ssh

This doesn't look intentional... Or its not a good idea if it is.




Reboot tells me WITNESS option is running - how did it get there
System comes up fine, gui ok... but as soon as i try to connect via CIFS ... same problem.

This goes for a while and then reboot again.

Its not really an issue, i am still testing stuff and dont have live data on the system.
I played around with tunables/sysctl values just before that (removing values for a 10gbe test and yes via gui not in files), maybe that caused an issue if the config files could not be written properly...

last reboot triggers another dump of interrupts again, but this time its fast at least.

Is it more sensible to try to recover or set up a new system ? I do have a config backup from last week, thats not too bad;)

 

[dlavigne]

What are the hardware specs?

 

[Rand]

E3-1230v3, 16gb ram
Had been running pretty good until the error after some initial issues;)

 

[Rand]

Well a reinstall and reimport of the config solved the issue.
If for some reason the broken installation is of interest - its still there for the moment.

 

[Rand]

So i've reinstalled the system on the old flash today;
after some initial issues with the hashes i got it to run, all fine after upgrading to 9.2.1.2.

At some point i ran 'poweroff';
next boot - same issue again, constant reboot, Witness option active, apparently a core log was written (saw it flashing by),
its starting tace/reboot right after "writing RSA key"

Can someone point me to documentation how to proceed?
I need to find what's causing this issue but not a clue how to get to the bottom of it...

Tried looking around in single user mode but couldnt find any logs or messages ...

 

[cyberjock]

What are the hardware specs?

E3-1230v3, 16gb ram
Had been running pretty good until the error after some initial issues;)

So a CPU and RAM is all that a computer is made up of! Wow.. I'm gonna go build a new system right now since I can save so much money by not buying a PSU, motherboard, etc.

 

[Rand]

I can save so much money by not buying a PSU, motherboard, etc.

You bought all that stuff? What for?

Anyhow while I only listed what i might call short specs you didnt answer my q how to get to debugging it at all;)
So here are the long specs

SC836
Intel S1200RPSv3
Intel Xeon E3 1230v3
16GB DDR3 Kingston (do you need the precise product number?)
Seasonic SS-300M1U
FreeNas installed on a 2GB eUSB Intel Flash
2x Dell H310 in IT Mode, LSI Phase 16
Toshiba 3TB x8
3xSSD400M (Testing)
2x600GB SAS HGST
Dell XR997 10GBe

 

[cyberjock]

Outstanding!

Take a look for your S1200 motherboard in the forums.. there's a few people that have had weird problems with it. One of those threads might be useful.

 

[Rand]

Havent seen anything that relates, unless i missed it. The system has been fine stability related up until 9.2.1.2, thats the first time i had those issues.
O/c it could be a coincidence and its a hardware issue after all, but i would need access to the logs or core files to check that, and i cant do that atm:(
I will try to boot a live cd later and mount the eUsb, thats probably the easiest way

 

[Rand]

Hm, found out that the step causing the issue is 'testing the samba configuration' ...

And could no space left on the Root device cause such a behavior or is that more effect on a small disc (logs, cores) then cause of it?

 

[Rand]

Ok ... there's something i dont understand...
According to the manual a 2GB Flash drive is sufficient.
I have just removed all partitions from the 2GB eUSB and installed 9.2.1.2 new, choosing "Install New", wipe partition.
Now i end up with a 1GB partition which is 97% in use *and all my config settings are still there*.

1. Wtf is my config still there? Does it autoimport it from the .system dataset (which still exists) ?
2. Why is it only using a 1GB partition for / when the drive has 2GB space?

I have 26972 bytes left... *Anything* writing to / (like that samba kernel dump i just deleted before me prevous post) would cause the disk to be full and possibly cause the panic.
And i saw an option flashing by turn_on_debugging_on_panic=1 ->1 so i guess thats whats happening here?

I mean easy workaround using a larger drive but before i go ahead i do that and create a bug re documentation i want to have somebody crosscheck my logic here please - and maybe explain why only 1GB is being used for / and what the other 1GB partition is doing?