Samba crashes frequently under load on FreeNAS-11.0-RC3 (1d8ee7925)

[ajschot]

Yeah this was a bad move on my part However the issue appeared in 9.10 U4 , In regards to the FreeNAS 11-RC4 I cannot see that on my update train I am using the FreeNas 11 stable train am I missing something?

i Also selected FreeNAS 11 Stable but i am still on RC3 it does not show me RC4 when i check.

 

[Crm]

I don't think any exactly mirror what he's describing. Especially since he upgrade to FN11 to try to fix the problem. If he can reproduce it on bare metal, I'll punt the issue to the bugtracker. :D

So an update, I installed 11 stable on a fresh VM rathen than 9 then upgrading. Same issue.

However when I looked on the bug tracker it mentioned the enable monitoring check box on the Directory tab and that it is experimental,. When I uncheck this I never saw the issue again and its been stable ever since, Even the bizarre permission behaviours I was getting are gone. So now going to try this fix on v9 or on the original VM to see if it was the fresh install the check box or both that fixed it, Stay tuned!

 

[gpsguy]

i Also selected FreeNAS 11 Stable but i am still on RC3 it does not show me RC4 when i check.

RC4 hasn't been released yet. I am sorry if I created some confusion by suggesting that one look for similar issues on the bug tracker.

That being said, it sounds like @Crm learned about a setting he had checked. I am curious to see how his testing goes.

 

[ajschot]

So an update, I installed 11 stable on a fresh VM rathen than 9 then upgrading. Same issue.

However when I looked on the bug tracker it mentioned the enable monitoring check box on the Directory tab and that it is experimental!

Hmmm i have the same issue where is that checkbox? i can not find a directory tab...??

 

[gpsguy]

Download an offline PDF of the 11.0 docs from 2trux.freenas.com and search for "enable monitoring". On page 159 there's a screenshot of the page showing the checkbox.

 

[Klontje]

Hello all, I am on bare metal running 11.0-RC3 and having the exact same issue and already applied the parameter shown in one of the first replies. I had the check set to 60 seconds (default) and had the error as noted by topic starter in the warning light with a timestamp of the last minute. Changing it to a longer period of time did not change the error time however.

 

[Crm]

Hello all, I am on bare metal running 11.0-RC3 and having the exact same issue and already applied the parameter shown in one of the first replies. I had the check set to 60 seconds (default) and had the error as noted by topic starter in the warning light with a timestamp of the last minute. Changing it to a longer period of time did not change the error time however.

Hi there here is the setting below highlighted in yellow. I had it checked and when I uncheck it the issue goes away. From what I can tell it appears that when the ixActiveDirectoryService attempts to heartbeat with the Domain Controller it fails and ends up cycling SAMBA & AD. Then it comes back but only to cut out again when the "AD Check Connectivity Frequency" value is reached. The max you can set it to is 3600

I can recreate the issue and it appears to clear up by not using the monitoring option.

Happy to do any other checks that anyone is looking for me to do if it helps, I can test this on a baremetal server to do to make sure its not VM related (Doubt it is mind you)

 

[Klontje]

I had disabled it but the errors didn't go away. Changing the interval didn't work for me neither. Strange thing is, I put 6TB of data on this setup by using robocopy over SMB; due to the bottleneck of my previous NAS (old Windows server 2012R2) it transfered at a max of 50MB/s. Now when I try to do a Synctoy operation from my desktop the synctoy operation is hanging in the indexing phase looking for differences. Transfering files using explorer maxes out the network connection and works fine.

Just rebooted with having the health check disabled; no warnings but synctoy still hangs in the indexing part looking for changes. Also browsing with explorer is not possible anymore until I wait about a minute; it looks like Samba is locking itself or something. I hope somebody will be able to analyze logs and get to the bottom of this...