11.3 Not Very Stable

[jlw52761]

I've been experiencing some very odd issues since upgrading to 11.3, and subsequently 11.3-U1, where the GUI just randomly hangs. I try to restart django and nginx, but after hours the service just will not stop, and I cannot successfully kill-9 the processes either. These leads me to shutting down the entire stack and rebooting FreeNAS.

On the same system, I deleted a dataset yesterday and the box randomly rebooted, without warning or any error/crashdump present that I can find. I never had this under 11.2, and I rebuilt by secondary system with 11.2-U8 and so far it's been rock solid, so I will start moving items over to this system to rebuild the existing primary box.

I'm wondering if anyone else is having these types of stability issues wince upgrading to 11.3 or if it's just me.

 

[Scharbag]

My 11.3-U1 install has been stable. The GUI can be a PITA at time - but a simple refresh solves any issues.

Cheers,

 

[hervon]

No big issue here on 11.3-U1. Only problem I had is when I rebuilt jails. Had to leave the interface blank otherwise I would get 'vnet failure'.

 

[Scharbag]

Oh, and I should be clear that I converted all of my jails to iocage in 11.2 so I did not have to deal with older warden jails anymore. Additionally, I do not use plugins as I create jails then install the applications I want in those jails. I am a strange duck.

FreeNAS has been good to me since V8.x :) What a long journey.

 

[ThreeDee]

no issues here ...

 

[pschatz100]

To the OP: Since you haven't provided your system specs, and are changing things at a low level, it is not possible to comment on your situation - if 11.2-U8 works for you, then stay with it. If you really want to move to 11.3, then read up on what has changed to see if anything would affect what you are doing.

11.3-U1 has been stable for the vast majority of users, and the outstanding issues are well documented. If you have problems with the GUI, then try clearing the cache on your browser to see if that helps. This has been commented on by many posters in the forum.

 

[jlw52761]

Main specs are
Dual Xeon(R) CPU E5-2603 0 @ 1.80GHz 4-core (No Hyperthreading)
128GB DDR-3 RAM
Intel S2600GZ Motherboard
Dual Intel I350 Gigabit Network Connection Quad Port
Not sure what low level things I’m changing unless managing a dataset is what you are considering low level.

Currently I am experiencing the issue where the GUI is non responsive and I cannot restart a service or kill the PID. The browser cache has been cleared, but that honestly doesn’t have an effect on trying to restart the nginx or django services and the processes just won’t die. Oh, and running something like lsof or top just hangs the console as well.

As you can imagine, since processes won’t die, the shutdown command does absolutely nothing and I have to power down the box the hard way, which since this is server class hardware I can do via Out Of Band.

so I do plan on going back to 11.2 once I move all the data to my secondary system. Maybe it’s just some driver in the newer version of FreeBSD that doesn’t like my hardware. I know the I350 NIC seems to be a sticking point for doing anything in the network like VLANs and such, so it’s entirely possible.

 

[jlw52761]

Unfortunately it also took NFS with it, which could be the buggy component. Anyone else running NFS with iSCSI?

 

[Yorick]

Just to tick that box and not discounting the possibility this is a bug in 11.3: Are you booting from USB, or SSD?

If this is a USB stick going bad on upgrade - not that uncommon - you might see something in the logs to that effect.

 

[SweetAndLow]

Main specs are
Dual Xeon(R) CPU E5-2603 0 @ 1.80GHz 4-core (No Hyperthreading)
128GB DDR-3 RAM
Intel S2600GZ Motherboard
Dual Intel I350 Gigabit Network Connection Quad Port
Not sure what low level things I’m changing unless managing a dataset is what you are considering low level.

Currently I am experiencing the issue where the GUI is non responsive and I cannot restart a service or kill the PID. The browser cache has been cleared, but that honestly doesn’t have an effect on trying to restart the nginx or django services and the processes just won’t die. Oh, and running something like lsof or top just hangs the console as well.

As you can imagine, since processes won’t die, the shutdown command does absolutely nothing and I have to power down the box the hard way, which since this is server class hardware I can do via Out Of Band.

so I do plan on going back to 11.2 once I move all the data to my secondary system. Maybe it’s just some driver in the newer version of FreeBSD that doesn’t like my hardware. I know the I350 NIC seems to be a sticking point for doing anything in the network like VLANs and such, so it’s entirely possible.

These issues seem a little excessive and probably not actually freenas being buggy but more like hardware or configuration.

I have my issues with 11.3 but nothing like this. For me it was shadow copies don't work with legacy snapshots, ipv6 auto conf in jail causes kernel panic and smb removing settings after the upgrade which broke smb shares. All of which i have bugs filed for and they are working. See the difference though all the things a called out as issues are very specific and have fixes acociated with them.

 

[ajschot]

Still issues with SMB and FTP.... but it seems that i am the onlyu one with these issues

 

[jlw52761]

@Yorick It is a USB stick with no signs of issues, but my second system running 11.2 U8 started doing the same thing when I migrated load over. I started really taking a look and noticed the autotune script added the kern.ipc.nmbclusters setting, which is also present on my 11.3 box. I've removed that from the 11.2 box and so far, under load, it seems stable. I'm wondering if that was the entire issue all along, as I've had other issues with the i350 NICs like not being able to use VLANs and such. Maybe a different NIC is in order...

 

[Yorick]

I think I remember reading that autotune is best avoided, as it may add parameters that are rather undesirable. This may be one such case.

I still expect your USB sticks to die. If not now, then eventually.

 

[jlw52761]

Already in process of re-working the box to have mirrored boot drives on a set of 5k 500GB spinners.