I’have a few questions – probably someone can help me in this (esx 3.5U2 and openfile 2.3).The iqn entered in the esx server (name change requires a reboot) must be the inititor name (and therefore different from the iqn of the iqn target name as defined in the openfiler – is this right?I have 2 esx-servers – for failover they should connect to the same storage. Can they connect to the same iqn or is it better to provide a secondary target iqn for the second esx?After some experiments with bonded/trunked ports it looks like my share was damaged (coud not reconnect). Are there any tools to check vmfs storages or to. Hoping someone can help me test OF 2.3 and esxi iscsi connection. Steps are below:1. Installed OF 2.3 as a vm on esxi (host#1).
![]()
Configuration Guide - Openfiler as Shared Storage for Virtual Environment Lab. Related Section: How to Install Openfiler to use as Shared Storage in Virtual. Openfiler appliance deployed as VM in Vmware Workstation. VLearning (5 ) Free Tools (1 ) vEvents (5 ) vEducation (3 ) What's New (2 ).
(Did not use appliance)2. Configured iScsi on OF and was able to map the lun.3. On a separate esxi (host#2), I’m trying to connect to the iScsi target.a. On host#2 I created a iscsi connection using vmkernel on a seperate port from the esxi host management networkb. My ip address for the iscsi connection is in the same subnet as my OF vm.c.
On the iSCSI Software Adapter (vmhba32) of host#2, I’ve enabled and configured the iScsi initiatord. On OF, the IP of the step(a) is allowed access to the target.questions:1. No luns appear after scan job. What have I missed? OF has an ip of 192.168.0.155my ESXi box has an ip of 192.168.0.127I added the ip for my esx box in OF and also added the ip access in the Network ACL list but I try to add the ip for OF in Virtual Infrastructure and it scans but does not find any available storage. I only have one nic so I cannot create a virtual switch for iscsi.I have both connected directly to an 8 port gigabit switch and my OF box has two nic’s that are teamed to the ip listed above.Any ideas on how to get this working?I am running ESXi off of a USB Flash drive but everything seems to be working I can get into Unsupported mode on ESX and it sees everything correctly it is just not seeing the iscsi storage on OF. Hi James,I have no real preference between the two of them TBH.
They are both really good servers and ideal for the ole home/work lab. I guess the ML115 G5 (AMD) has the advantage over the ML110 G5 (Intel) at the moment as it offers a quad core CPU – as opposed to the ML110’s dual core.I do like the Shuttle XPC’s because of their small size. I only used one for my filer as I had one lying around that used be my main PC a few years ago.OpenFiler is pretty flexible when it comes to the hardware it’ll run on and doesn’t require a particularly large amount of CPU or memory resource.
A decent 1Gb Ethernet adapter is highly desirable though.If you have some old kit kicking around you could most likely get it running on that.Failing that EBay has some cheap older shuttles for sale quite often. My shuttle is a Shuttle XPCSK21(G10) AMD model.Let us know how you get on. OF works great with ESXi 4.0, make sure you get the 2.3.1 “respin” which has 2.6.26 kernel as this installs better on modern hardware. I skip all that network access stuff and just create the iSCSI lun (as above, but with write-back and fileio) and then map it (as above).
Saves time because the ACL was jsut another barrier to getting vmotion and vstorage motion working. With ESX/i 4.0 3x I/O performance improvement and 10x iSCSI performance improvement (and finally support for jumbo frames) makes sense to pretend ESX3.5 never existed. Ok, so I easily got iSCSI up and running with these great instructions. Well except for one problem – I’d like to have two LUNs. One for a 2TB volume and the other for a 1TB volumes.I’ve tried just mapping the two LUNs to the one target without luck and then tried creating two targets and mapping one to each. No matter what I do I can’t get the second LUN to show up in the iSCSI Software Adapter.I’m using ESX 3.5.4 and Virtual Center 2.5.4. I’ve tried editing the initiators.deny file as suggested without luck.Does anyone have any ideas or is this something that isn’t supported by Openfiler?Thanks.
Never mind, I got it working by removing all the configuration after a full upgrade and reboot and doing it all again.Then I found out about VMWare LUN size limit. My 3TB LUN only showed as a much smaller storage location.
So, I undid the larger 3TB volume and broke it into two separate 1.5TB volumes. Then mapped those two to the target.I now have three LUNs – 1TB, 1.5TB and 1.5TB.Once all my ESX servers could see it I then added a VMFS 2 storage locations. The second one I added the second 1.5TB LUN to as an extent.Now I’m fully in business. Thanks for these great instructions. Thanks for the reply Simon.We made significant progress yesterday/todayin that we managed to get the box back up and on the network with all the services available.We made use of command line config files to delete any/all references to the new switch/virtual net(esxcfg-nics/esxcfg-vmknic/esxcfg-vswitch, for everybody’s info. These files are located under /sbin).Once we deleted the references, we rebooted the box, but we were stumped/gutted for a minute when nothing was working. Turned out that when we went into the network/nic card administration, the nic card was not enable.
Checked that and we were suckin diesel again.Thnanks again Simon. Hi Jason,I would recommend looking at splitting your iSCSI (OpenFiler) traffic off onto a seperate subnet and if using the same physical switch as the other traffic (eg: VM traffic) then look at using VLAN tagging. Although this is not necessary it will help keep things seperate.I’m not saying that this is cause of the issue you mention as if (correct me if I’ve misread this) you put your iSCSI port on the 192.168.0.x network you can connect your ESX host to your iSCSI based LUN presented from your OpenFiler server?If this is the case then your OpenFiler server has it’s network port (assuming it just has one) configured to the wrong subnet (ie: 192.168.0.x and not 192.168.50.x). How I think it should look:Build in NIC – ESX Management – 192.168.0.x3 x NICs – 192.168.50.x – VM Traffic (or perhaps 2 x NICs for VM Traffic and then 1 x NIC for vMotion)1 x NIC – 192.168.60.x (new subnet) – iSCSI traffic. Ensure that the OpenFiler iSCSI port is also configured to use an IP on this new subnet.
You could always use one of the 4 NIC ports you’ve got as a failover NIC for iSCSI traffic – though you’d really want seperate physical switches to get the most benefit from this resilient configuration.Hope this makes sense and that I read your message ok.Cheers,Simon. Simon,Excellant documentation. I managed to figure out how to do it myself before I found this article but like the guy above, I can find the iSCSI LUNs when I use my main Vswitch on the two ESX3i servers (172.25.120 mask 255.255.0.0) but I have a separate Dual Gigabit NIC in both ESX3i servers and also in the OF Appliance. I setup one interface in the OF as 192.169.0.69 and the other as 192.168.1.69 and one ESX3i server has a NIC setup as 192.168.0.6 and the other ESX3i as 192.168.1.4 with direct cable from Openfile from each ESX3i server to the Openfiler on the relevant NIC. I can Ping each ESX3i server from the Openfiler console but when I try and discover the LUN after changing the Target in each ESX3i box to the relevant 192.168.0.69 or 192.168.1.69, it fails to find the LUN but if I change it back to 172.25.200.69 it finds the LUN straight away. Hi Simon,First off, thanks for you posts. I’ve learned much because of what you write.
Thank you thank you thank you.Now, I was reading through your openfiler setup with VMware, and i was wondering you already run into this issue: I have 4 ESX host. I can only connect 3 host to openfiler using iscsi. Everytime I connect the 4th one, one of the connected 3 will lose its connection to isci, and if I look into the vmkernel logs, it would display something like:Sep 17 17:27:17 esx22 vmkernel: 0:07:24:41.099 cpu0:4194)WARNING: NMP: nmpDeviceAttemptFailover: Retry world failover device “t10.F405E5F6D26603C677” – issuing command 0x410004034e40I thought the solution was to change the MAXConnection setting in openfiler, but the default setting is “1” and I can successfully connect 3so I’m a bit lost.Hope you can shed some light into this. GREAT GUIDE!!!!That is what I have been looking for a while, found parts and bits but not a very complete guide that focus on the confusing part of the OpenFiler process (Volumes).I hope the OpenFiler folks hire you to create their next guide.
П™‚I have a question regarding the subnetmask.I have a lab network and I use 192.168.1.x ip address range with 255.255.255.0 subnetmask all my VM and ESX hosts communicate with no problem between eachother but I read in your forum that people had problem making OpenFiler see the ESX server when they use this subnetmask and had to change to 255.255.255.255 to work. Will I need to change mine subnetmask so my ESX server can see openfiler?Thanks for your asssitance.Neo. HelloOne more question.I did not find yet in any guide that would give me the answer to my question below:Scenario:I have Windows 7 running Vmware workstation 6.5 I created inside of the Vmworkstation 2 ESX 3.5 (VMs) on it.
I want to install OpenFiler and make the 2 ESX servers (vms) see the shared Storage so I can test VMotion, DRS, HA features.I downloaded the iso image for OpenFiler 2.3 x86 and I want to install OpenFiler through my Vmware Workstation.Since there is no instructions in how to do this I believe you would need first to create a NEW VM option on my Vmware Workstation 6.5.There I selected linux 4 64 bit and so forth. And selected install OS through the DVD (my openfiler iso I burned on a DVD).My QUESTION is during the VM creation when it gets the time to select disk space size it usually shows 8 gigs (default). But since I’m creating Openfiler and I want to have at least 50 gigs of disk space to be shared and used by the 2 ESX hosts.Is it here on this step where I would change from 8 gig to 50 gigs and then after OpenFIler is installed that I would go inside and configured the 50 Gig diskspace into shared storage?The above step I did not find anywhere yet. Hi there,My shuttle has two disks – though I have recently moved my OpenFiler to a ML110 G4 as I can fit up to 4 drives in it.You certainly can use OpenFiler with one disk. The ESX/ESXi hosts you connect to the shared disk that the OpenFiler presents don’t mind whether this disk is in a RAID configuration or not. All they (the ESX/ESXi hosts) care about is that they have some shared disk and from there you can test HA, DRS, FT (CPU permitting).
Of course, as you no doubt know, by not having RAID you are not protected from physical disk failure and coule lose everything if this was to happen.Hope this helps.Cheers,Simon. Cheers for the quick reply.At present I have bought 2 ML115s to use as the ESX(i) Cluster. I am not sure If I can get access to a 3rd machine that has more than 1 disk drive (hence my question). Ideally I would like a 3rd OpenFiler box with 2 disks to use RAID, but may have to settle for a single disk, proof of concept, machine to begin with (maybe just an old desktop, if I can get hold of one).I may have a HP XW4300 that I can use as the OpenFiler at a later date and I think this can hold 4 disks etc, although (physically) it is bigger than a ML115!! Hi’ I’m a newbie at networking and I’m having trouble setting up HA in my network for one of my class. The problem is that I have 2 ESX server box, and a SAN box.
The goal was to put all my vms on my SAN box and then connect with the ISCI connect to one of my servers, and then create a cluster with the other one, so that if one goes down, I have the other one to back it up. The problem also is that when I trying to connect the storage adapter to my esx box, it doesn’t show up on my server. I was wondering if I should create a trunking port connection from my SAN box to my switch so that it can communicate with any of my VLANS. Hi There,I wondered if you could help me? I have recently moved our vCenter install to a more stable server, removed the license from our existing vmhost (running ESX 4.0),disconnected the host and uninstalled vCenter. Installed vCenter on to another server, installed fine, no problems, added our new host, also running esx 4.0, with no errors, but when I tried to add our existing host, vCenter gave me the following error:“Cannot install the vCenter agent service. Not enough space on device.”After I got this i found out how to uninstall the vCenter agent from the host by running the following commands:1.
Established if the Virtual Centre agents are installed by running this command ‘rpm -qa grep vpxa’ and I got this reply ‘VMware-vpxa-4.0.0-162856’.2. I then stopped all the services by running the following:service mgmt-vmware stop && service vmware-vpxa stop && service vmware-vmkauthd stop && service xinetd restart3. I removed the Virtual Centre agent by running ‘rpm -ef VMware-vcpa-4.0.0-162856’4.
I then removed the ‘vpxuser’ by running ‘userdel vpxuser’5. Restarted the services by running:service mgmt-vmware start && service vmware-vmkauthd start6.
I waited a few minutes and then ran ‘vmware-cmd -l’ to make sure the services had restarted.Now once the services had restarted, I tried to added the host again but it gave me a different error:“Cannot install the vCenter Server agent service. Unknown installer error”Any ideas as to what might be wrong? I would be grateful for any help at all!!Many thanks,James. Thank you for posting this guide.
![]()
I am running OF 3.2 on a DL385G1, ESXi 4.1 on a DL360G5 and am now able to access my iSCSI datastore under ESXi. It turns out I needed these comments to get the job done, however, as I needed to comment out the line in /etc/initiators.deny as Peter Barfield mentions.There were a few other spots where I stubbed my toe:1) Did not catch how far down I’d have to scroll, in some instances, in the OF config web pages.2) I’d left my floppy drive enabled in BIOS on the DL385 (OF box) which caused i/o errors to occur continually on /dev/fd0, which in turn kept me from being able to do much in the web console to configure OF. Disabling it was all it took to calm things down.3) Authentication within OF proved to slow me down a bit too – read two different guides from other sources. One of them incorrectly advised me to include the FQDN for a NETBIOS name (duh!
I should have known better), which put it over the limit for name length.FMR:1) Best practices?2) VLAN configuration, considerations?3) Performance tuning?Again, great work.Gus. Hello Simon,Very nice article; Here is my situation and see if you can help me out. I have two physical servers with 8GB of RAM on one and 4GB of RAM on the other. I tried Oracle VM and Oracle Virtual Box both unsuccessful and a lot of times wasted. These two physical servers (not VMS) both have RHEL 5 with update 5 installed on them. I am a Senior Database Analyst and am in the process of getting my OCE (Oracle Certified Export) certification for Grid Infrastructure for Oracle 11g R2.
![]()
Hi,Great article and helps a lot of people including myself. Hi,Thank you for this article.would you please help me with the following once i created the whole stuff(mentioned above):1. How can i use this space in a Virtual machine on the same network (whose ip i mentioned in network access conf menu)?2. Do i need to map this drive into my remote machine if so what would be the protocol (as in windows we write?3. If this storage space gets automaticaly mapped into te remote machines which were mentioned in net-access menu what would be the mount location by default??kind regards:Gireesh.
![]() Comments are closed.
|
AuthorWrite something about yourself. No need to be fancy, just an overview. Archives
March 2023
Categories |