toasters

toasters@lists.teaparty.net

2 participants
13529 discussions

VIF Issues
by Nail, Thomas E 16 Nov '12

16 Nov '12

All, I am needing to re-order some of the virtual interfaces on an HA pair of 3170s running OnTap 7.3.7. The current configuration has: 2 x 1 Gbe interfaces in an LACP 'vif_r0' 2 x 1 Gbe interfaces in an LACP 'vif_r1' 1 second-level vif, vif_rsec, comprised of vif_r0 and vif_r1. What I want to (eventually) get to is: 4 x 1 Gbe interfaces in an LACP vif_r0 And I don't want to take a downtime to do it. Unfortunately, with an HA pair, this isn't going to be possible (I don't think…).Once a failover and configuration change is made, the takeover partner doesn't have a "correct" interface to give back to, since it's "partner" interface name has changed. So what I'm proposing is: Leave the second-level vif alone. Remove the physical interfaces from vif_r1 and put them in vif_r0. The question I have is: can I fully de-populate a vif of physical interfaces? Will it stick on the last one? I don't have anywhere to test this, so I'm hoping someone with some spare capacity can either test it and let me know if it will work. Thanks in advance, -=Tom Nail

2 1

how significant / non-optimal path error
by Stuart Kendrick 13 Nov '12

13 Nov '12

How substantially can this affect performance? scsitarget.partnerPath.misconfigured:error]: FCP Partner Path Misconfigured - Host I/O access through a non-primary and non-optimal path was detected. We have a two-node FAS3020 cluster running 7.2.4, serving two clients via Fibre Channel. [More precisely, four clients: two pairs of Active/Standby SQL Server clusters. A pair of Fibre Channel switches provide the fabric for this tiny SAN.] Cleverly, we've configured the clients to cross the cluster interconnect in order to reach their shelves. i.e. under normal operation, Client1's traffic must traverse the fabric, ToasterA, the cluster interconnect, and ToasterB, before reaching its shelves. Conversely, Client2's traffic must traverse the fabric, ToasterB, the cluster interconnect, and ToasterA before reaching its shelves. Occasionally, we see the error message above in the logs ... sometimes just once/week, occasionally twice/day. We'd like to ignore this ... we hope to retire the installation sometime next year ... but on the other hand, if this degrades performance substantially, then we'll reconsider. ? --sk Stuart Kendrick FHCRC

2 1

no thread since 10/31/2012
by Uddhav Regmi 12 Nov '12

12 Nov '12

Just making sure... From: toasters-bounces(a)teaparty.net [mailto:toasters-bounces@teaparty.net] On Behalf Of Sto RageC Sent: Wednesday, October 31, 2012 11:18 PM To: Ray Van Dolson Cc: toasters(a)teaparty.net Subject: Re: Clarification on CIFS Permissions (ONTAP 8.0.2P3) On Wed, Oct 31, 2012 at 5:45 PM, Ray Van Dolson <rvandolson(a)esri.com> wrote: Where is this default "Everyone" permission set set by default? It doesn't seem to be tied to the CIFS share security settings so I presume it's an NTFS qtree default? Thanks, Ray Yes, the NTFS security defaults to Everyone when a NTFS qtree is created. We normally don't set the permissions at the share level, it is set at the security level on NTFS. The first step is to remove Everyone, add a share administrator - full control and then the user groups with modify. -HTH

2 1

Looking for Source
by Carolyn Chambers 10 Nov '12

10 Nov '12

Hello, I am looking for source code for the old symlink_tool.exe tool. I understand that it is not supported but it is very handy to have around. I have an idea for its next life... Thanks, Carolyn ________________________________ Carolyn Chambers 972-679-4211 - cell 972-767-0555 - fax 972-578-2708 - office cc(a)storageassessments.com<mailto:cc@storageassessments.com> 1255 W. 15th Street, Suite 825 - Plano, TX 75075 [cid:image001.png@01CDBE9F.F83542D0]

1 0

probably a n00b quota question
by Randy Rue 06 Nov '12

06 Nov '12

Hello All, Found an anomaly with how a quota is being applied on a vFiler and it feels like I must be missing something obvious. We have a vFiler that serves up a couple of hundred folders for researchers. This is on 7.3.5 so the aggregate is limited to 16TB. We've set a 500GB quota per folder (qtree), strictly to hopefully keep any one researcher from tanking the filer. Here's the /etc/quotas file: #Auto-generated by setup Fri Dec 18 11:16:33 PST 2009 * tree@/vol/cs_researcher 500G - 450G - - /vol/cs_researcher/Rue_R tree 13G - 10G - - /vol/cs_researcher/tosh_M tree 1024G - 950G - - If I read this correctly, all qtrees are limited to 500G with two exceptions, me (testing) and one researcher that we allowed to go to a TB. But if I look at the volume I have one folder with 6TB in it. But if I run a "quota report" on that qtree, it says they're using 437GB: cs@tungsten-a> quota report /vol/cs_researcher/Bradley_P K-Bytes Files Type ID Volume Tree Used Limit Used Limit Quota Specifier ----- -------- -------- -------- --------- --------- ------- ------- --------------- tree 16 cs_researcher dley_P 458297068 524288000 3261880 - /vol/cs_researcher/dley_P It's worth noting that a Windows reporting tool (TreeSize) shows the folder has 350K folders and 11.3M files but the above shows 3.2M files. In either case, will that overload the system's ability to track usage? And why would the two tools show two numbers? Also, while the well-behaved folders are primarily touched via NTFS/CIFS and show AD owners, the tricky folder is mostly NFS and shows as owned by root. But while we often hit quota situations on our home drive filer (user based quotas) where garbled ownership mixes up quotas, in this case (qtrees) it shouldn't make any difference? Am I missing something? Hope to hear from you, Randy

1 0

Clarification on CIFS Permissions (ONTAP 8.0.2P3)
by Ray Van Dolson 01 Nov '12

01 Nov '12

Have a volume with an NTFS qtree being shared out via CIFS. The CIFS share has a number of different share level permissions. When I browse to the share from a Windows client and view Security permissions on a folder within, I see only the user "Everyone" listed with implicit full control. Where is this default "Everyone" permission set set by default? It doesn't seem to be tied to the CIFS share security settings so I presume it's an NTFS qtree default? Thanks, Ray

2 1

vFiler with vlan tagging?
by Paul Letta 25 Oct '12

25 Oct '12

Hello, I'm trying to deploy multiple vFilers on a single filer. The vFilers need to be on different vlans. I want to use the same VIF (made up of 2 interfaces) for this. Am I wrong in my thinking that I can use vlan tagging with these multiple vfilers while using the same vif? I think I should be able to use the default IPspace on the vif. I've had our network people trunk 2 vlans to the ports that make up the VIF. I then create 2 vFilers, each specifying its own IP address/subnet mask. (The IPs and masks are different for each vfiler). But what I'm missing is the connection between vfiler and vlan. With ESX, we trunk dozens of vlans to the nics that the esx server is using for the vswitch. We then configure the vm's to be on which ever vlan we need it to be. Of course that's ESX's virtual switch doing the work for me. How does this picture map into vfiler / physical filer space? Thanks, Paul

4 3

Re: vFiler with vlan tagging?
by Scott 25 Oct '12

25 Oct '12

First I created the vfilers and issued them front end IP addresses, aliased to the front end VIF on the controller. Then, I created a back end VIF for hosts to mount their storage (e0b & e1b). On that VIF I created 4 vlans. Once the vlans were created, I went back to the vfiler and added a back end IP, then aliased that IP to the particular vlan. Of course, the networking guy did his magic on the switch side creating the vlans. E: flags=0xa2f48863<UP,BROADCAST,RUNNING,MULTICAST,TCPCKSUM,VLAN> mtu 1500 ether 02:a0:98:0f:01:79 (Enabled interface groups) E-34: flags=0x2b48863<UP,BROADCAST,RUNNING,MULTICAST,TCPCKSUM> mtu 1500 inet 10.54.34.6 netmask 0xffffff00 broadcast 10.54.34.255 ether 02:a0:98:0f:01:79 (Enabled interface groups) E-36: flags=0x2b48863<UP,BROADCAST,RUNNING,MULTICAST,TCPCKSUM> mtu 1500 inet 10.55.10.6 netmask 0xffffff00 broadcast 10.55.10.255 inet 10.55.10.220 netmask 0xffffff00 broadcast 10.55.10.255 ether 02:a0:98:0f:01:79 (Enabled interface groups) E-38: flags=0x2b48863<UP,BROADCAST,RUNNING,MULTICAST,TCPCKSUM> mtu 1500 inet 10.53.10.6 netmask 0xffffff00 broadcast 10.53.10.255 ether 02:a0:98:0f:01:79 (Enabled interface groups) E-40: flags=0x2b48863<UP,BROADCAST,RUNNING,MULTICAST,TCPCKSUM> mtu 1500 inet 10.56.10.6 netmask 0xffffff00 broadcast 10.56.10.255 ether 02:a0:98:0f:01:79 (Enabled interface groups) On Oct 25, 2012, at 4:33 PM, Coartney, Paul wrote: > How did you create the vfilers? Can you send me the output of > > Ifconfig -a > >> From vfiler0. > > On 10/25/12 1:24 PM, "Christopher S Eno" <s.eno(a)me.com> wrote: > >> In our case, I created a VIF of two physical interfaces, then assigned 4 >> different vlans to that VIF. Then, the vfiler's IP addresses were >> aliased to whichever vlan they needed. Using default IP space. >> >> >> On Oct 25, 2012, at 4:13 PM, Paul Letta wrote: >> >>> Hello, >>> >>> I'm trying to deploy multiple vFilers on a single filer. The vFilers >>> need to be on different vlans. >>> I want to use the same VIF (made up of 2 interfaces) for this. Am I >>> wrong in my thinking that I can >>> use vlan tagging with these multiple vfilers while using the same vif? >>> >>> I think I should be able to use the default IPspace on the vif. >>> >>> I've had our network people trunk 2 vlans to the ports that make up the >>> VIF. >>> I then create 2 vFilers, each specifying its own IP address/subnet mask. >>> (The IPs and masks are different for each vfiler). >>> >>> But what I'm missing is the connection between vfiler and vlan. >>> >>> With ESX, we trunk dozens of vlans to the nics that the esx server is >>> using for the vswitch. >>> We then configure the vm's to be on which ever vlan we need it to be. >>> Of course that's ESX's >>> virtual switch doing the work for me. >>> >>> How does this picture map into vfiler / physical filer space? >>> >>> >>> Thanks, >>> Paul >>> >>> >>> >>> >>> _______________________________________________ >>> Toasters mailing list >>> Toasters(a)teaparty.net >>> http://www.teaparty.net/mailman/listinfo/toasters >> >> _______________________________________________ >> Toasters mailing list >> Toasters(a)teaparty.net >> http://www.teaparty.net/mailman/listinfo/toasters >

1 0

test an upgrade without fully failing back an HA node?
by Randy Rue 24 Oct '12

24 Oct '12

Hello All, We just had a rough night trying to upgrade a 3020 from 7.2.4 to 7.3.7 where the upgraded node came up with no working network connections. Turns out the same setup / RC file that works great in 7.2.4 crumps in 7.3.7. After some flailing, we used revert_to to roll back. In the mean time, our VMware farm had several hundred VMs suffer a ~1 hour loss of their disks and ~40 VMs needed help ranging from just a reboot to a full rollback of the disk image. This Sunday we're scheduled to upgrade a v3170 pair that supports about three times as many VMs and uses the same network design. We have some reasons to hope that the same issue won't bite us, and also have better options to test before we have to risk the virtual farm, and some other ideas on how to mitigate risk and impact. But it was also suggested that an even better way to avoid a problem would be if we could upgrade a node and then test the RC file before it was actually "given back." If I do a takeover from Node B, then upgrade Node A, is there a way to bring up Node A far enough to test that its RC file loaded successfully but before it tries to go back into production? Hope to hear from you, Randy

7 7

Max CIFS Multiplex
by Ian Dotson 23 Oct '12

23 Oct '12

The online "Health Check" was telling us that: sam.misconfig.badValue.cifsMaxMpx: Max Multiplex should not be near or equal to the value of the option cifs.max_mpx. The value of cifs.max_mpx should be set higher. Please discuss what values are acceptable for this parameter with a NetApp technical support representative. Then, use "options set cifs.max_mpx ..." to set this option to a new value. We're seeing unusually large high-water mark for cifs max multiplex -- up to 511 (from 'cifs stats -c') We do have some client IIS servers (Server 2003, IIS 6.0) that we suspect are the culprits. "cifs stat -h <ip>" doesn't display the max number of multiplexed cifs sessions requested for a host, only current stats. Is there a way to determine which client had asked for this huge number of multiplexed cifs sessions? Thanks, Ian Dotson

3 2

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

toasters