dekkar
asked on
vSphere 4 and Storage / Luns
Hello, I just ran into an issue on my vsphere setup...
I have a SAN (iscsi) with multiple Luns on it, I have also setup a separate Windows based SAN (for testing and backup)
and have added this to vsphere via iscsi.
All works well, and as expected. The problem is, when I take one lun offline, vms on other luns go offline. Or they kind of go offline, then back online and cycle through being on and off.
At first, I thought this was because all the luns were on the same SAN..... but today I restarted the windows iscsi server, hence taking its lun offline, and all of a sudden, all my VMs were in this half disconneted state...
They remained disconnected, where I couldnt ping them or access them in any way, until the windows iSCSI server had restarted, then everything came back on line and everything was up and running OK
The weird thing about this, is the windows iscsi server has no VMs running on it.... all the VMs are running on the SAN which holds all my production luns....
Anyone know why it seems if I take one lun offline, all luns become unresponsive?
I have a SAN (iscsi) with multiple Luns on it, I have also setup a separate Windows based SAN (for testing and backup)
and have added this to vsphere via iscsi.
All works well, and as expected. The problem is, when I take one lun offline, vms on other luns go offline. Or they kind of go offline, then back online and cycle through being on and off.
At first, I thought this was because all the luns were on the same SAN..... but today I restarted the windows iscsi server, hence taking its lun offline, and all of a sudden, all my VMs were in this half disconneted state...
They remained disconnected, where I couldnt ping them or access them in any way, until the windows iSCSI server had restarted, then everything came back on line and everything was up and running OK
The weird thing about this, is the windows iscsi server has no VMs running on it.... all the VMs are running on the SAN which holds all my production luns....
Anyone know why it seems if I take one lun offline, all luns become unresponsive?
ASKER
hmmmm not too sure if I understand...
We have a equallogic san..... around 6 LUNs, each with a VM running on it. I have a 2008 server with starwind installed hosting a LUN
They are all connected to vsphere with the vmhba33:C0:Txx:L0 runtime name. (xx = lun number I presume)
This is including the LUN hosted on the windows box.
We have a equallogic san..... around 6 LUNs, each with a VM running on it. I have a 2008 server with starwind installed hosting a LUN
They are all connected to vsphere with the vmhba33:C0:Txx:L0 runtime name. (xx = lun number I presume)
This is including the LUN hosted on the windows box.
Ok, I get thet setup now.
What happens when you take the windows iscsi san offline and rescan the controllers? Do the LUNs for the Dell SAN come up or stay down?
Another way to test is to login to each esx host directly and see if what you're seeing is from vCenter or from ESX host.
What happens when you take the windows iscsi san offline and rescan the controllers? Do the LUNs for the Dell SAN come up or stay down?
Another way to test is to login to each esx host directly and see if what you're seeing is from vCenter or from ESX host.
ASKER
I havent really had a chance to test it..... I run ping sessions to a few of the VMs running, and when I restarted the Windows server, all VMs pings dropped out, I couldnt RDP, and the vcenter vm also because unresponsive, hence vcenter disconnected.
This sorted itself out once the windows server was back up and its lun available.
I have had this problem before when I 'unpresented' a lun on the equallogic box, which I had already deleted according datastore from both ESXi boxes..... But I thought possibly it was because all the luns were on the same SAN, and same IP address etc etc, so ESX had a meltdown and thought everything on that IP was disconnected.
I am a little hesitant to test this, as I dont know if this pausing of VMs will risk corrupting them......
This sorted itself out once the windows server was back up and its lun available.
I have had this problem before when I 'unpresented' a lun on the equallogic box, which I had already deleted according datastore from both ESXi boxes..... But I thought possibly it was because all the luns were on the same SAN, and same IP address etc etc, so ESX had a meltdown and thought everything on that IP was disconnected.
I am a little hesitant to test this, as I dont know if this pausing of VMs will risk corrupting them......
Sounds like vCenter is losing connectivity but most likely the ESXi hosts are ok. Are you running Update 1 of ESXi and is your vCenter virtual or physical?
Also, This windows server isn't the DNS server for the VM's is it?? or more so, the SQL server for vCentre??
Regards
Krystian
Regards
Krystian
ASKER
we havent updated vsphere to 4.xxx update.
and vCenter is virtual.
No, the DNS server is physical. vCenter and its SQL database are on same VM.
and vCenter is virtual.
No, the DNS server is physical. vCenter and its SQL database are on same VM.
The issue is most likely due to vCenter being a VM. When the storage controller do a rescan vCenter loses connection and thus you lose connectivity to it is what I'm thinking.
How are the hosts added to vCenter via IP address or FQDN?
How are the hosts added to vCenter via IP address or FQDN?
ASKER
IP address..... Which is what the solution provider setup.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Hmmmm I did prepare the network for ESX and FQDN, as this is what I thougth was supposed to happen. But he told me it doesnt make much difference...
and who was I to question a VMware expert ;)
ok, willl look into changing this.
So I need to remove each ESX box from vSphere, and re-add them via their FQDN? The actual VMs running on the ESX boxes wont really be affected yes? as vSphere only takes care of vmotion etc etc?
and who was I to question a VMware expert ;)
ok, willl look into changing this.
So I need to remove each ESX box from vSphere, and re-add them via their FQDN? The actual VMs running on the ESX boxes wont really be affected yes? as vSphere only takes care of vmotion etc etc?
That's correct if you already have the hostname/domain on the hosts than you can disconnect and remove. Then add back in via hosname. This is integral part of vCenter, HA, etc.
Per VMware Best Practice
kb.vmware.com/kb/1009080
-------------------------- ---------- ----
Make sure that the system you use for your vCenter Server installation belongs to a domain, rather than a workgroup.
It is critical that you have reliable DNS and Time services.
During the installation, the connection between the machine and the domain controller must be working.
Log into the system using an account with local administrator rights. If joining another vCenter Server in Linked Mode, the account must be a local Administrator on both systems.
The computer name cannot be more than 15 characters.
Assign a static IP address and host name to the Windows server that will host the vCenter Server system. This IP address must have a valid (internal) DNS registration that resolves properly from all managed ESX hosts.
An old KB, but shows that you should have the ESX hosts added via FQDN, either DNS (recommended) or host file on vCenter server
http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1005750
Per VMware Best Practice
kb.vmware.com/kb/1009080
--------------------------
Make sure that the system you use for your vCenter Server installation belongs to a domain, rather than a workgroup.
It is critical that you have reliable DNS and Time services.
During the installation, the connection between the machine and the domain controller must be working.
Log into the system using an account with local administrator rights. If joining another vCenter Server in Linked Mode, the account must be a local Administrator on both systems.
The computer name cannot be more than 15 characters.
Assign a static IP address and host name to the Windows server that will host the vCenter Server system. This IP address must have a valid (internal) DNS registration that resolves properly from all managed ESX hosts.
An old KB, but shows that you should have the ESX hosts added via FQDN, either DNS (recommended) or host file on vCenter server
http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1005750
ASKER
OK, will give it a go.
Let me know if it works..or not. We'll troubleshoot further if needed
Create a target for Windows and a Target for vSphere or at least zone the luns via iSCSI initiator so that only Windows can see the Windows LUN and vSphere can only see vSphere LUNs
What type of SAN are you using?