Tunables & Fatal Error

MediaManRIT

Dabbler
Joined
Jul 14, 2013
Messages
17
I am trying to set some sysctl values using Tunables. I got two in. On the third, the page's blue-spinner just keeps spinning. If I try adding a third, I get a message in red "Fatal error! Check logs.". But, no files are updated in /var/log when that happens. Nothing new in middlewared.log or messages.

I'm on FreeNAS-11.3-U4.1 .

Anyone have any guesses?

Thanks!
 

Mlovelace

Guru
Joined
Aug 19, 2014
Messages
1,111
I am trying to set some sysctl values using Tunables. I got two in. On the third, the page's blue-spinner just keeps spinning. If I try adding a third, I get a message in red "Fatal error! Check logs.". But, no files are updated in /var/log when that happens. Nothing new in middlewared.log or messages.

I'm on FreeNAS-11.3-U4.1 .

Anyone have any guesses?

Thanks!
What tunable are you trying to add and it's value?
 

MediaManRIT

Dabbler
Joined
Jul 14, 2013
Messages
17
What tunable are you trying to add and it's value?

I was attempting (not well) to tune for 10Gbps. There in lies my failure I believe. I set these before the lockout happened:

net.inet.tcp.recvbuf_inc=2097152
net.inet.tcp.recvbuf_max=16777216
net.inet.tcp.recvspace=4194304

I just rebooted, and most things couldn't get socket space. So I manually reverted those three back to default values in the shell (at the console, using sysctl command), then I was able to start sshd and get remote access again.

Now I'm trying to figure out how my custom values that I put in the UI are set when the system boots, so I can remove them and get back to baseline, since in the UI it still hangs even after restarting middlewared.
 

MediaManRIT

Dabbler
Joined
Jul 14, 2013
Messages
17
Figured it out. Django wasn't running. Once I started that, I was able to go into the web ui and change the values.
 

Mlovelace

Guru
Joined
Aug 19, 2014
Messages
1,111
I was attempting (not well) to tune for 10Gbps. There in lies my failure I believe. I set these before the lockout happened:

net.inet.tcp.recvbuf_inc=2097152
net.inet.tcp.recvbuf_max=16777216
net.inet.tcp.recvspace=4194304

I just rebooted, and most things couldn't get socket space. So I manually reverted those three back to default values in the shell (at the console, using sysctl command), then I was able to start sshd and get remote access again.

Now I'm trying to figure out how my custom values that I put in the UI are set when the system boots, so I can remove them and get back to baseline, since in the UI it still hangs even after restarting middlewared.
These look fairly reasonable, what are your system's specs? What 10Gbe card are you running?
 

MediaManRIT

Dabbler
Joined
Jul 14, 2013
Messages
17
These look fairly reasonable, what are your system's specs? What 10Gbe card are you running?
X520 NIC. 32GB RAM. I think I found the true root cause. Order of Operations failure. I set those three before I set kern.ipc.maxsockbuf . I think that was the fail. When I rebooted with just those 3 set, nothing that needed a socket was able to start. Once I manually bumped maxsockbuf I was able to start services.
 
Top