9.2 to 9.2.1.1 upgrade, continuous reboots

[Raj Rajagopal]

I upgraded my FreeNas from 9.2 to 9.2.1.1 using the web UI method.

Everything seemed to go through fine but after the reboot (during which I see couple of warning relating to 'failed precmd routine for vmware_guest' and 'unable to start syslogd'. After these messages, either the system continues to boot fine or awaits at the loading kernel modules for a really long time (15+ minutes).

Finally, the boot sequence completes and I see the freenas home menu (options 1-11). I can access the web UI at this point. But, the system automatically restarts within 5 to 10 minutes. This has been going on every since my upgrade and I can't find anything else in the var/log/messages (possibly due to syslogd down?).

Any help on debugging this? I also have my 9.2 DB backed up, should I just revert back 9.2 or can I do a clean install 9.2..1.1 and import the DB?

Thanks,

 

[Yatti420]

What kind of hardware? Is watchdog enabled?

 

[Raj Rajagopal]

simple, straightforward hardware - AMD 965 Black Edition running on GA-780G-UD3H gigabyte motherboard and everything connected using native motherboard features. 4 disks in ZFS raid on sata and motherboard LAN.

I noticed another similar error : https://bugs.freenas.org/issues/4150. So, disabled the watchdog (no option in the bios, go went the rc.conf route) but still the same issue.

 

[dlavigne]

can I do a clean install 9.2..1.1 and import the DB?

Yes, let us know if that fixes the issue.

 

[Raj Rajagopal]

Yes, let us know if that fixes the issue.

I ran in to mDNSResponder errors, when I tried a clean 9.2.1.1.

So, gave up and went back to 9.2. Everything is fine now. Next tweaking is setting up OpenVPN client.

will revisit 9.2.1.1 or later revision upgrade, sometime in the future.

 

[Vadim Nemtsev]

Hello. I have the same issue. Had to go back to 9.2.
Running Freenas on Intel Nas ss4000e with 2GB of RAM and E2200 Pentium dual core 4x3TB HDD Raid-Z. I gave a try clean and gui upgrade but both failed. The machine is headless what make it difficult to resolve or diagnose the problem.

 

[cyberjock]

Hello. I have the same issue. Had to go back to 9.2.
Running Freenas on Intel Nas ss4000e with 2GB of RAM and E2200 Pentium dual core 4x3TB HDD Raid-Z. I gave a try clean and gui upgrade but both failed. The machine is headless what make it difficult to resolve or diagnose the problem.

You realize you are using 1/4 of the minimum RAM for ZFS? Don't be surprised when your pool suddenly stops working someday.

Good luck!

 

[Vadim Nemtsev]

@cyberjock
Thanks for your recommendation, however with dedup disabled it`s enough to have 2GB, in fact i`m running this server since 8.0.1 with no issues at 99.99% uptime (from install to upgrade), and it`s capable of running 4 jails with Transmission, Plex, MiniDLNA, and BT sync, NFS, cifs/Samba, FTP. And yes, i`m using enterprise grade drives on my small home multimedia NAS.
And the last, your comment do not answer the question why newer firmware wont accept db backup from absolutely stable and working system.
Good luck!

 

[cyberjock]

And you didn't get the subtle hint that not meeting the minimum requirements may be responsible?

I hate to break it to you, but basically none of us help people with problems that don't meet the minimum system requirements.

 

[Vadim Nemtsev]

And you didn't get the subtle hint that not meeting the minimum requirements may be responsible?

I hate to break it to you, but basically none of us help people with problems that don't meet the minimum system requirements.

Probably you're right, it's possible cause. At weekend i"ll try to figure out the problem and will post the results of my investigation. Please do not take it as personal disrespect, i really appreciate your work and i follow up your topics on this forum, but the responsibility of minimum system requirements in my case under big doubt. IMHO of course.

 

[cyberjock]

Yeah.. I can't tell you how many problems that are random and unexplained suddenly disappear when the minimum specs are met.

For that reason I don't even consider troubleshooting problems if the minimum specs aren't met. It's just not worth my time when we already know better than to expect stability from not meeting them. The 8GB of RAM is for stability and not performance.

 

[Raj Rajagopal]

Yeah.. I can't tell you how many problems that are random and unexplained suddenly disappear when the minimum specs are met.

For that reason I don't even consider troubleshooting problems if the minimum specs aren't met. It's just not worth my time when we already know better than to expect stability from not meeting them. The 8GB of RAM is for stability and not performance.

While I agree with your general comments, my setup had 16GB of RAM (more than enough to meet the min requirements). it was (and is) working fine in 9.2 and the problem appeared after upgrade to 9.2.1.1. I don't think it's entirely due to specs and in my opinion, some parameter (either in the system or sqlite) gets messed up to cause the reboots.

 

[cyberjock]

While I agree with your general comments, my setup had 16GB of RAM (more than enough to meet the min requirements). it was (and is) working fine in 9.2 and the problem appeared after upgrade to 9.2.1.1. I don't think it's entirely due to specs and in my opinion, some parameter (either in the system or sqlite) gets messed up to cause the reboots.

Sure. But look at what I said...

Yeah.. I can't tell you how many problems that are random and unexplained suddenly disappear when the minimum specs are met.

For that reason I don't even consider troubleshooting problems if the minimum specs aren't met. It's just not worth my time when we already know better than to expect stability from not meeting them. The 8GB of RAM is for stability and not performance.

Gotta start ruling out stuff somewhere. And the first and easiest thing to rule out is meeting the minimum specs. If the software won't even execute because of out of RAM problems and whatnot caused by not meeting the minimum requirements, you're going to pull your hair out long before you get to troubleshooting code.

Do you have any idea how pissed off you'll be when you try to start the CIFS service 50 times in a day over an 8 hour binge trying to get FreeNAS to work and 10 of those times it'll start and the other 40 it won't and you'll have no freakin' clue what the hell is going on? Yeah, that's what people without the minimum specs often see. You'll try to find patterns and you'll wonder what the hell you are doing wrong, because you'll figure it's user error. But if you change no settings and keep clicking start it might start once out of 10. So then you think FreeNAS is a shitty software package that is horribly unstable because if you keep clicking it, it will eventually start. No, it's actually the fact that you decided to ignore the minimums and are looking for excuses for your own negligence. ;) I've seen it here PLENTY of times.

And I didn't say it was for certain the the fact that he had so little RAM. All I said was that it wasn't worth *my* time to try to help him if I can't even guarantee that if a service won't start it *isn't* because of how little RAM he has. After all, if you don't have enough RAM the error message doesn't say "not enough RAM". It says the same exact thing it says if you have one of about a dozen other errors. So you have to rule out the "stupid simple" stuff first.

It's called "minimum specs" for a reason. You don't want to meet our minimum requirements, that's fine. Don't expect our help. It's not much different than showing up here without providing minimum information as requested by the forum rules when you have a problem. You'll see your thread ignored, deleted, or you get banned. We have minimum standards that we expect when you post.