Freenas v8.3.1 - NFS Crashes Server

Status
Not open for further replies.

Kobus0834

Cadet
Joined
May 29, 2013
Messages
2
Hi, I have a single server with 35TB volume. I have created a Iscsi share 25 TB and wanted to add 2 x 5TB NFS shares. I added the shares and about 15-20 minutes later my server froze up. I had to hard boot the server and disable the NFS. Below is the error messag I received before rebooting the server.

Error:
Fatal trap12: Page fault while in kernel mode
Fault Virtual Address = 0x80c71284
Fault code = supervisor read data, page not present
Stopped at vaccess_acl_nfs4+0x86: cmpl %ecx,0x4(%rdx)
Thank you

Server Spec:Intel(R) Xeon(TM) CPU 2.66GHz
12GB RAM
FreeNAS-8.3.1-RELEASE-p2-x64
 

jgreco

Resident Grinch
Joined
May 29, 2011
Messages
18,680
Re: Iscsi and NFS - one Volume Same server

Please read and comply with the rules or you may be temporarily banned.

You can edit your post and add/fix content or the Title.
Posts not meeting requirements and not updated may be deleted in 24 hours or sooner (Rule #4).

http://forums.freenas.org/showthread.php?1-Forum-Rules

woo a Canned Reply! :smile: But seriously, since it should work, you've given us nothing to work with.
 

Kobus0834

Cadet
Joined
May 29, 2013
Messages
2
Re: Iscsi and NFS - one Volume Same server

Please read and comply with the rules or you may be temporarily banned.

You can edit your post and add/fix content or the Title.
Posts not meeting requirements and not updated may be deleted in 24 hours or sooner (Rule #4).

http://forums.freenas.org/showthread.php?1-Forum-Rules

woo a Canned Reply! :) But seriously, since it should work, you've given us nothing to work with.

Hi jgreco,

Did you have a chance to look at my problem as yet after I have added the necessary information?

Thank you!
Kobus
 

jgreco

Resident Grinch
Joined
May 29, 2011
Messages
18,680
Didn't notice because I don't think it throws posts back into the "New" bin if you merely edit them. And it's not particularly thorough to describe a server by just naming the CPU speed and RAM.

But that's sufficient to make the following observation: that seems consistent with the sorts of panics people see when the system is stressed for memory.


For a 35TB volume, expect to need at least double the RAM (24GB) you have. There's a plausible argument to be made that the recommendation is actually for 8GB *plus* 1GB per TB, so on the slightly paranoid side one could even argue that you should be up at 48GB of RAM.

So, having framed things that way, comments:

1) There's no inherent reason that a smaller memory configuration cannot be made to work. I've got a 12x4TB array on the bench that runs under ESXi which means I can twiddle how much memory it gets. It can be made to work on 8GB but performance kind of tanks.

2) iSCSI seems to be somewhat less intensive on ZFS and kernel than NFS. I suspect that may help to explain why you were getting away with running iSCSI on such a large pool with small memory. I've run FreeNAS with ZFS for iSCSI on 1GB and I think even less, though not for anything important, and it required tuning.
 
Status
Not open for further replies.
Top