Kern.maxfiles?

[Milkwerm]

Hi
I'm running the most reecnt 9.2.1.2 release (post errata version) and this morning it was completely unresponsive after 2 days of running fine.
the console was constantly scrolling this:
kern.maxfiles limit exceeded by uid 00, please see tuning(7)

I had to reboot the box to get access back up, but is there anything I could/should be doing to try and prevent this happening again?

 

[Yatti420]

What are the specs of your NAS box? Are you using autotuning or any systctls? http://forums.freenas.org/index.php?threads/kern-maxfiles-limit-exceeded-by-uid-0.17971/

Are you running MiniDLNA or any plugins that access files alot?

 

[Milkwerm]

Hi Yatti
The server is a intel white box. S5000PAL mainboard, Xeon E5520 CPU, 12G Ram and 5x 1TB in RaidZ config. (yes. quite old. It's been happily running freenas since 8.2) I've just replaced the USB stick (8G sandisk)
It's using AD auth and only serving CIFs shares. no auto-tuning turned on. and unmodified from install in anyway other than to set up the windows sharing, join the domain and enable SSH.

 

[Milkwerm]

I'm also seeing a lot of this now in the console:
Mar 5 11:03:02 NAS01 smbd[28544]: Failed to check packet auth. (NT_STATUS_ACCESS_DENIED)
Mar 5 11:07:16 NAS01 kernel: pid 28213 (smbd), uid 21201: exited on signal 11
Mar 5 11:23:15 NAS01 smbd[4180]: [2014/03/05 11:23:15.067964, 0] ../source3/smbd/server.c:556(smbd_accept_connection)
Mar 5 11:23:15 NAS01 smbd[4180]: accept: Software caused connection abort
Mar 5 11:33:30 NAS01 smbd[23888]: [2014/03/05 11:33:30.564649, 0] ../source3/smbd/close.c:1181(close_directory)
Mar 5 11:33:30 NAS01 smbd[23888]: Could not close dir! fname=<file path>, fd=-1, err=9=Bad file descriptor
Mar 5 11:33:30 NAS01 kernel: pid 23888 (smbd), uid 27057: exited on signal 11

And TOP has a large amount of RUN and KQREAD states for samba..

Anyone?

 

[cyberjock]

I'd love to give you an answer, but I'm not sure I have one to give. My recommendation is to make an account and open a ticket at bugs.freenas.org. Or, drop into IRC and message me and I'll see if I can hustle a dev into looking at your problem. 6pm-midnight in the USA is usually a good time to be in IRC if you want support.

 

[Milkwerm]

Thanks Cyberjock. I'm in New Zealand so need to check out the time zone difference :)
Are there any logs in particular you would recommend attaching to a bug report?

 

[cyberjock]

No, I'd simply specify your hardware specs, what services you use, and the error you get. The devs will probably ask for more or may even know what the problem is or that it's already patched in the nightlies.

 

[Milkwerm]

Bug report created #4474

 

[Yatti420]

SMB/CIFS Issue?

 

[Milkwerm]

Would certainly be my guess Yatti420 :).

 

[Glen Baars]

I can confirm this on two NAS servers that I have.

Services are : SMB, Active directory, SSH, NFS, Smart

[root@stored] ~# sysctl kern.maxfiles kern.maxfilesperproc kern.openfiles
kern.maxfiles: 12328
kern.maxfilesperproc: 11095
kern.openfiles: 11107
fstat is reporting lots of these:
root smbd 93944 86 /mnt/DATA/ 129530 -rwxrwx--- 36300800 r
root smbd 93944 87 /mnt/DATA/ 76195 -rwxrwx--- 512 r
root smbd 93944 88 /mnt/DATA/ 86083 -rwxrwx--- 34 r
root smbd 93944 89 /mnt/DATA/ 186707 -rwxrwx--- 100327424 r
root smbd 93944 90 /mnt/DATA/ 86082 -rwxrwx--- 36947456 r
Hardware is supermicro x8 with xeon X3430 and 16GB ecc memory.
Controller details:: Model 9650SE-16ML, 16 ports, Firmware FE9X 4.10.00.026, BIOS BE9X 4.08.00.004
16 x hitachi enterprise 4tb drives with 3 x 5 drive raidz1 + hotspare
We where running a previous version fine with no changes expect upgraded via gui.
These are domain intergrated, low load boxes ( hold backup images )
Any further info is available from glen at back u p s . n e t . au
I've got 30+ freenas boxes like this! Over 1.5PB of drives!
PS: Cyberjock - thanks for all the good info on the forum, you have helpled me alot !
Glen