ifconfig socket / no buffer space errors

[dknm]

OMG IT broke! Of course I did nothing! ;)

Ok, so I'm scratching my head with this one.

I'm running 11.1-U6 under ESXI 6.5 for a number of months with decent uptimes (1-2 months unbroken). Now I can only access the shell.

From a cold boot it 1st reboots itself after the zfs volume's imported


on the second boot it keeps spewing ifconfig: socket (family 2, sock_dgram): no buffer space errors. It won't bind itself or jails to preset IPs - so that's no good.



Last thing I did was add a large (500,000+) image library into Plex, which must've generated thumbnails that ate the space. *sigh* I'm now aware there's a swap file that's been turfed out.




However, the boot drive looks fine to me:

Expanding the boot drive size from 8 to 32gb in ESXI didn't register - I'd need to alter the partition table somehow. Is that the way to approach it?



In here the partition scheme shows ada0p3 as the swap, whereas mine doesn't exist?!



Could you point me in the right direction please?

 

[dknm]

I figured it out. The error in question has to do with the tunables configuration - not the swap file. I realised at some point, I added some lines to fix a transmission error that kept popping in the console

Code:

transmission-daemon[16338]: UDP Failed to set receive buffer: No buffer space available (tr-udp.c:65)

I don't understand why it took until yesterday to cascade into the errors above, but I had to reset the configuration. I'll leave it here, maybe it'll help someone.

You will need console access - IPMI / ESXi will do.

The system backs the configuration nightly in

Code:

/var/db/system

inside you'll find a configs-xxxxxx folder (the xxx is a string specific to your build). Inside, there are folders for each of the 11.1 upgrades. I was after the "FreeNAS-11.1-U6 (caffd76fa)” folder.

I copied a few files over to where my pool storage is (/mnt/Data/Share). I would advise to copy more than you think you need, as once I restored the config - I could no longer see the nightly backups anymore.

With this done, I exited and reset the configuration to defaults (option 8).

I skipped the initial wizard, imported the volume, setup samba to access the backups I copied into /mnt/Data/Share and uploaded it in System-General

That wasn't fun.

 

[dknm]

I'm back to step one. Before any of the socket errors, I had swap_pager_getswapspace failed errors that made everything unresponsive. And I had one just earlier:



Swap use is... negligible


Help.

//edit - updated plex from 1.12 to 1.14. Things stabilised for now.