Guest LDOM disk access through multiple IO domains

Hi All,
I am working on a configuration, wherein the boot disk to the Guest LDOM is being provided through an image file hosted on a VxVM diskgroup (vmdg1). The configuration has another copy of the same image file being provided through another VxVM diskgroup (vmdg2)l through another virtual disk service.
let me clear the configuration a little more in detail
A T-5240 server having 2 IO domains configured
Primary (Control domain + IO domain + Service domain) configuration
A VxVM diskgrop (vmdg1) having a boot image file
Secondary (IO domain + Service domain) configuration
A VxVM diskgroup (vmdg2) having a copy of the boot image file
these devices are exported through their respective virtual disk services with the same mpgroup name to a guest LDOM. The vdsisk is then assigned to the guest ldom which is using the volume through the primary service.
When the guest LDOM is started it starts with the disk export through the Primary domain. All writes happen fine. When the VxVM diskgroup is deported from the Primary the Guest LDOM still remains online as starts using the disk image path through the secondary domain.
I then bring the VxVM diskgroup and the mounts back online on the Primary domain and deport the diskgroup from the secondary domain to see if it failsback to the image through the Primary domain. The Guest LDOM now is in a hung state and does not allow access through the local console or through network logins.
Has anyone see such a problem? Also is it recommended to use disk based image as a backend device in mpgroups through "ldm add-vdsdev" ?
TIA,
Sudhir

As far as I know then only way to "re-balance" the I/O across the domains is to unbind/bind the guests. Not a great answer, but this could be done as part of the guests patching cycle.
I think there is an RFE to provide MPxIO-like features to guests.

Similar Messages

Hyper V guest slow disk access

I have a Hyper V cluster(2012) with a 2012 guest running 2012 acting as my file server.
My 2 hosts are connected to a gig switch stack which in turn is connected to my Dell MD3200i.
My file server has a "data" disk that is a VHD(fixed) on the iSCSI storage. As is it OS(dynamic)
I'm coping 200GB right now and only seeing data thoughtput of 8MB/sec.
When I check perfrmance on the file server the disk queue is huge. But when I check the queue for C: on the host, which is where the cluster storage appears to sit, its got a low if not no queue.
Also did a copy using the host and saw 58MB/sec so I'm happy the network should be able to pass the data.
Switch CPU is at 5%
Can anyone tell me what I'm doing wrong?

Fixed VHD always performs better than Dynamic VHD in most scenarios by roughly 10% to 15% within the exception of 4K writes, where fixed VHD performs significantly better according to the Windows Server Performance Team Blog.
Hyper-V and VHD Performance - Dynamic vs. Fixed:
http://blogs.technet.com/b/winserverperformance/archive/2008/09/19/hyper-v-and-vhd-performance-dynamic-vs-fixed.aspx
Witch doesn't match your tests, so I think you missed something in your configuration schema. Read the following guideline:
Performance Tuning Guidelines for previous versions of Windows Server:
http://msdn.microsoft.com/en-us/library/windows/hardware/dn529134
Regards,
If you found this post helpful, please give it a "Helpful" vote. If it answered your question, remember to mark it as an "Answer". This posting is provided "AS IS" with no warranties and confers no rights! Always test ANY
suggestion in a test environment before implementing!

Error "NOTICE: [0] disk access failed" during guest domain network booting

Hi,
Could you please tell me what is the problem with my configuration?
I created guest domain on my T1000 server.
As a disk I used disk from disk array: /dev/dsk/c0t18d0
I added disk using commands:
# ldm add-vdsdev /dev/dsk/c0t18d0 vol1@primary-vds0
# ldm add-vdisk vdisk1 vol1@primary-vds0 myldom1
# ldm set-variable auto-boot\?=false myldom1
# ldm set-variable boot-device=/virtual-devices@100/channel-devices@200/disk@0 myldom1
Then I logged to guest domain and booted from network to install OS from JumpStart server:
{0} ok boot net - install
Boot device: /virtual-devices@100/channel-devices@200/network@0 File and args: - install
Requesting Internet Address for 0:14:4f:f9:78:19
SunOS Release 5.10 Version Generic_137137-09 64-bit
Copyright 1983-2008 Sun Microsystems, Inc. All rights reserved.
Use is subject to license terms.
Configuring devices.
NOTICE: [0] disk access failed.
Checking rules.ok file...
Using begin script: install_begin
Using finish script: patch_finish
Executing SolStart preinstall phase...
Executing begin script "install_begin"...
Begin script install_begin execution completed.
ERROR: No disks found
- Check to make sure disks are cabled and powered up
Solaris installation program exited.
Configuration:
[root@gt1000a /]# ldm list-bindings
NAME STATE FLAGS CONS VCPU MEMORY UTIL UPTIME
primary active -n-cv- SP 4 2G 0.5% 2h 23m
MAC
00:14:4f:9f:71:4e
HOSTID
0x849f714e
VCPU
VID PID UTIL STRAND
0 0 5.3% 100%
1 1 0.5% 100%
2 2 0.5% 100%
3 3 0.4% 100%
MAU
ID CPUSET
0 (0, 1, 2, 3)
MEMORY
RA PA SIZE
0x8000000 0x8000000 2G
VARIABLES
keyboard-layout=US-English
IO
DEVICE PSEUDONYM OPTIONS
pci@780 bus_a
pci@7c0 bus_b
VCC
NAME PORT-RANGE
primary-vcc0 5000-5100
CLIENT PORT
myldom1@primary-vcc0 5000
VSW
NAME MAC NET-DEV DEVICE DEFAULT-VLAN-ID PVID VID MODE
primary-vsw0 00:14:4f:fa:ca:94 bge0 switch@0 1 1
PEER MAC PVID VID
vnet0@myldom1 00:14:4f:f9:78:19 1
VDS
NAME VOLUME OPTIONS MPGROUP DEVICE
primary-vds0 vol1 /dev/dsk/c0t18d0
CLIENT VOLUME
vdisk1@myldom1 vol1
VCONS
NAME SERVICE PORT
SP
NAME STATE FLAGS CONS VCPU MEMORY UTIL UPTIME
myldom1 active -n---- 5000 12 2G 0.1% 2h 18m
MAC
00:14:4f:f9:e7:ae
HOSTID
0x84f9e7ae
VCPU
VID PID UTIL STRAND
0 4 0.5% 100%
1 5 0.0% 100%
2 6 0.0% 100%
3 7 0.0% 100%
4 8 0.0% 100%
5 9 0.0% 100%
6 10 0.0% 100%
7 11 0.0% 100%
8 12 0.0% 100%
9 13 0.0% 100%
10 14 0.0% 100%
11 15 0.0% 100%
MEMORY
RA PA SIZE
0x8000000 0x88000000 2G
VARIABLES
auto-boot?=false
boot-device=/virtual-devices@100/channel-devices@200/disk@0
NETWORK
NAME SERVICE DEVICE MAC MODE PVID VID
vnet0 primary-vsw0@primary network@0 00:14:4f:f9:78:19 1
PEER MAC MODE PVID VID
primary-vsw0@primary 00:14:4f:fa:ca:94 1
DISK
NAME VOLUME TOUT DEVICE SERVER MPGROUP
vdisk1 vol1@primary-vds0 disk@0 primary
VCONS
NAME SERVICE PORT
myldom1 primary-vcc0@primary 5000
[root@gt1000a /]#
Kind regards,
Daniel

Issue solved.
There was a wrong disk name:
primary-vds0 vol1 /dev/dsk/c0t18d0
I changed to c0t18d0s2 and now I sucessfuly installed OS from Jumpstart.

How to provide access to multiple users connected to a Dumb switch? (multi-auth/multi-domain)

Good morning everybody,
I am writing on behalf of not being able to implement a desired outcome in our company network. In fact the situation is as follows:
What I want to do is to be able to authenticate users (802.1x authentication) in our company radius server and authorize them access by having a dynamic VLAN assignment in a multi-user environment on one and the same port of a Cisco 2960 switch. So far, the authentication and authorization has been working completely smoothly (there are no problems with itself). The concept involves the configuration of both DATA and VOICE VLANs as I there is also phone authentication implemented. In order to simulate this environment I introduce a Dumb switch connected to my Cisco 2960 Catalyst.
What I have successfully managed to get to work so far is this:
1) On one switch port I have tried the “authentication host-mode multi-domain” and it worked perfectly for a PC behind a telephone, or with one PC connected to a the dumb switch + the telephone connected to another port of the dumb switch. Logically it is the same situation as there is a separation in two domains – DATA and VOICE. Bellow is an output from show authentication sessions for this scenario.
Interface MAC Address Method Domain Status Session ID
Fa0/23 0021.9b62.b79b dot1x DATA Authz Success C0A8FF69000000F3008E (user1)
Fa0/23 0015.655c.b912 dot1x VOICE Authz Success C0A8FF69000000F9009F (phone)
2) On the other hand, when I try the same scenario with the “authentication host-mode multi-auth”, the switch still separates the traffic in two domains and is able to authenticate all users, AS LONG AS they are in the same VLAN.
show authentication sessions:
Interface MAC Address Method Domain Status Session ID
Fa0/23 0021.9b62.b79b dot1x DATA Authz Success C0A8FF69000000F3008E (user1)
Fa0/23 b888.e3eb.ebac dot1x DATA Authz Success C0A8FF69000000F8008C (user2)
Fa0/23 0015.655c.b912 dot1x VOICE Authz Success C0A8FF69000000F9009F (phone)
However, I cannot succeed authentication of many users from DIFFERENT VLANs, neither in multi-auth nor in multi-domain modes.
What I want to get is an output like this:
Interface MAC Address Method Domain Status Session ID
Fa0/23 0021.9b62.b79b dot1x DATA Authz Success C0A8FF69000000F3008E (user1)
Fa0/23 b888.e3eb.ebac dot1x DATA Authz Success C0A8FF69000000F8008C (user2)
Fa0/23 0015.655c.b912 dot1x VOICE Authz Success C0A8FF69000000F9009F (phone)
I want the switch to authenticate the users anytime they connect to itself and for them to have an instant access to the network. (I tell this because I tried scenario 1) with multi-domain mode and authentication violation replace, and it worked but, two users never had access to the “Internet” simultaneously!!!
The configuration of the interface connected to the Dumb switch is as follows.
interface FastEthernet0/x
description Connection to DUMBswitch
switchport mode access
switchport voice vlan XXX
switchport port-security maximum 10
switchport port-security
switchport port-security violation protect
authentication host-mode multi-auth
authentication priority dot1x
authentication port-control auto
authentication timer reauthenticate 4000
authentication violation replace
dot1x pae authenticator
dot1x timeout tx-period 10
spanning-tree portfast
The way I see it is explained in the following steps:
- PC1 connects to the Dumb switch. This causes the Cisco switch to authenticate user1. This creates an auth. session with its MAC address linked to a domain DATA.
- When PC2 connects to the Dumb switch, this causes the violation replace which replaces the recent authenticated MAC address with the MAC of PC2. I would like it once authenticated to appear in the authentication sessions with a link to a new DATA domain linked to the VLAN assigned from the RADIUS server.
Is this possible? I think (in theory) this is the only way to provide authenticated access to multiple users connecting through Dumb switch to the network.
Has anybody ever succeeded in such a configuration example and if yes, I would be love to get some help in doing so?
Thank you
Stoimen Hristov

Hi Stoimen,
I have done a setup similar to yours with the only exception being VLAN assignment. When I used dACLs only, it makes things somewhat easier as the VLAN no longer matters. Remember that the switchport is in access mode and will only allow a single VLAN across it (with the exception of the voice VLAN). I think that is the real cause of your problem.
From what I can see, you have 2 options available to you:
1) Use dACLs instead of VLAN assignment. This means that an access list will be downloaded from the radius server straight to the authenticated user's session. I have tested this and it works perfectly. Just Google Cisco IBNS quick reference guide and look for the section that deals with Low Impact mode.
2) Get rid of the dumb switches and use managed switches throughout your network. Dumb switches will always be a point of weakness in your network because they have no intelligence to do advanced security features like port security, 802.1x, DHCP snooping, etc.
Hopefully someone else will chime in with another option.
Xavier

Raw LUN access from guest LDOM

Hi,
Is it possible to have raw access to LUNs from guest LDOM ?
Scenario is - 2 mpxio HBAs conected to control domain, 20 LUNs presented, 10 for ZFS filesystems inside LDOMs and 10 with raw access for HORCM operations inside LDOMs for switching TrueCopy pairs. So each LDOm should have exclusive access to two LUNs. This is requred for campus cluster with TrueCopy replicated storage for HA service running inside two ldoms.
We do not have T4 connected to storage yet and I can't check output of ldm list-io, but I've yet to see any examples of such setup on the net so a bit worrying if this possible at all.
Thanks !
Edited by: 907565 on 12-Jan-2012 00:31

You might be looking at PCIe Direct I/O - This extends current PCIe support to enable you to assign either individual PCIe cards or entire PCIe buses to a guest domain.
You would probably need more pci cards on different slots

Can I use the new Time Capsule to backup my mid 2010 Macbook Pro? Also can I want to free up my hard disk, can I save my photos and files on the time capsule and later access through wifi?

Can I use the new Time Capsule to backup my mid 2010 Macbook Pro? Also can I want to free up my hard disk, can I save my photos and files on the time capsule and later access through wifi?

Can I use the new Time Capsule to backup my mid 2010 Macbook Pro?
Yes, if you are asking about using Time Machine to backup the Mac.
Also can I want to free up my hard disk, can I save my photos and files on the time capsule and later access through wifi?
You are not thinking of deleting the photos and files on your Mac, are you? If you do this, you will have no backups for those files.
Another concern is that Time Machine backs up the changes on your Mac. At some point, Time Machine will automatically delete the photos and files from the Time Capsule.....you just don't know when this might occur.
In other words, only delete files from your Mac that you can afford to lose.

Is there any way to recover files when the back up catalog has been deleted accidently. I can see a preview of the photos, but I cannot access or edit them. They say "the folder cannot be located" and "the photo is missing". I have gone through multipl

I can see a preview of the photos, but I cannot access or edit them. They say "the folder cannot be located" and "the photo is missing". I have gone through multiple Adobe tutorials to try and recover them to no avail. Any tips?

Backup catalogs wouldn't have helped with this problem anyway, but naturally, you shouldn't delete backup catalogs for a while.
The solution to reconnecting the folders is given here: Adobe Lightroom - Find moved or missing files and folders

Disk Access Failed while Installing Solaris Container.

I have setup 5 guest domains together with the Control Domain.
$ ldm list
Name State Flags Cons VCPU Memory Util Uptime
primary active -t-cv SP 4 4G 0.8% 2d 2h 15m
secondary active -t--v 5000 4 2G 0.5% 3h 5m
dmz active -t--- 5001 8 2G 0.0% 46m
sunray inactive -----
application inactive -----
identity active -t--- 5002 4 4G 0.1% 1h 2m
In each of the guest domain, I plan to install a number of Solaris Containers to run different applications. While installing one Solaris container (zoneadm �z <zone name> install) in dmz domain, I start the installation/configuration (either zoneadm �z <zone name> install or zlogin �C <zone name>) of another Solaris container in identity domain. Everything starts OK until half way I get an error message as follow:
Jun 13 18:33:16 dmz vdc: NOTICE: [1] disk access failed.
The installation of the Solaris Container in dmz halts as soon as I see the message above. Nothing will work except a force stop in the Control Domain using ldm stop �f dmz and restart.
What happen to the Solaris Container installation/configuration in identity domain? It either continues the process without error or fails with the same error message shortly after. For example,
Jun 13 18:33:24 identity vdc: NOTICE: [1] disk access failed.
The vdisks for each guest domain are setup by following the steps on section �Using ZFS Over a Virtual Disk� in LDoms 1.0 Administration Guide. Each guest domain is booted from a disk image in the Control Domain and has two 9GB data vdisks (mapped to two physical 9GB disks on A5200 disk array) running ZFS striping. Each guest domain vdisks are serviced by a separate vdisk server.
Any idea?

I am not sure if multiple virtual disk servers are officially supported under current LDoms release.
I had tried using only one virtual disk server. The same problem exhibited. I thought may be the vds was not able to keep up with the virtual disk I/O. That's why I setup multiple vds.

Lync 2013 certificate requirements for multiple SIP domains

Hi All,
I am engaged with a client in respect of a Lync 2013 implementation initially as a conferencing platform with a view to enabling EV functions (inc. PSTN conferencing) in the future. They initially need to support 30 SIP domains and eventually
around 100 SIP domains which is proving to be either not possible or severely cost prohibitive. Their current certificate provider, Thawte, can only support up to 25 SANs and have quoted them 5 figures. We tend to use GeoTrust as they are cheaper but they
appear to have a limit of 25 SANs. GoDaddy appear to support up to 100 SANs for a pretty reasonable cost. My questions are as follows:
Is there a way that I’m missing of reducing the number of SANs required on the Edge server?
Use aliases for access edge FQDNs - Supported by desktop client but not by other devices so not really workable
Don’t support XMPP federation therefore removing the need for domain name FQDNs for each SIP domain
Is there a way that I’m missing of reducing the number of SANs required on the Reverse Proxy server?
Friendly URL option 3 from this page:
http://technet.microsoft.com/en-us/library/gg398287.aspx
Client auto-configuration:
i.
Don’t support mobile client auto-configuration in which case no lyncdiscover.sipdomain1.com DNS records or SANs would be required.
ii.
Support mobile client auto-configuration over HTTP only in which case CNAME records are required for each SIP domain (lyncdiscover.sipdomain1.com, etc. pointing to lyncdiscover.designateddomain.com) but no SANs are required.
iii.
Support mobile client auto-configuration over HTTPS in which case DNS records are required for each SIP domain and a SAN entry for each SIP domains is also required. This is because a DNS CNAME to another domain is not supported over
HTTPS.
If the answer to 1 and/or 2 is no, are there certificate providers that support over 100 SANs?
How do certificate requirements differ when using the Lync 2013 hosting pack? I would think that this issue is something that a hosting provider would need to overcome.
Would the Lync 2013 Hosting Pack work for this customer? The customer uses SPLA licensing so I think is eligible to use the hosting pack but not 100% sure it will work in their environment given that client connections are supposed
to all come through the Edge where their tenants will be internal and also given the requirement for an ACP for PSTN conferencing.
Many thanks,

Many thanks for the response.
I was already planning to use option 3 from the below page for simple URLs to cut down on SAN requirement.
http://technet.microsoft.com/en-us/library/gg398287.aspx
What are the security concerns for publishing autodiscover over port 80? I.e. Is this only used for the initial download of the discovery record and then HTTPS is used for authentication? This seems to be the case from the following note on the below page:
http://technet.microsoft.com/en-gb/library/hh690030.aspx
Mobile device clients do not support multiple Secure Sockets Layer (SSL) certificates from different domains. Therefore, CNAME redirection to different domains is not supported over HTTPS. For example, a DNS CNAME record for lyncdiscover.contoso.com that redirects
to an address of director.contoso.net is not supported over HTTPS.
In such a topology, a mobile device client needs to use HTTP for the first request, so that the CNAME redirection is resolved over HTTP. Subsequent requests then use HTTPS. To support this scenario, you need to configure your reverse proxy with a web publishing
rule for port 80 (HTTP).
For details, see "To create a web publishing rule for port 80" in Configuring the Reverse Proxy for Mobility. CNAME redirection to the same domain is supported over HTTPS. In this case, the destination domain's certificate covers the originating
domain.”
I don’t think SRV records for additional SIP domain access edge is a workable solution as this is not supported by some devices.
As per the below article:
http://blog.schertz.name/2012/07/lync-edge-server-best-practices/
“The recommended approach for external client Automatic Sign-In when supporting multiple SIP domains is to include a unique Access Edge FQDN for each domain name in the SAN field. This is no longer a requirement (it was in OCS) as it is possible to
create a DNS Service Locator Record (SRV) for each additional SIP domain yet have them all point back to the same original FQDN for the Access Edge service (e.g. sip.mslync.net).
This approach will trigger a security alert in Windows Lync clients which can be accepted by the user, but some other clients and devices are unable to connect when the Automatic Sign-In process returns a pair of SRV and Host (A) records which do not share
the same domain namespace. Thus it is still best practice to define a unique FQDN for each additional SIP domain and include that hostname in the external Edge certificate’s SAN field”.
===================
1. Basically the requirement is to initially provide Lync conferencing services (minus PSTN conferencing) to internal, external, federated and anonymous participants with a view to providing PSTN conferencing and therefore enterprise voice services later.
2. The customer currently supports close to 100 SMTP domains and wants to align their SIP domains with these existing domains. The structure of their business is such that “XXX IT Services” provide the IT infrastructure for a collection of companies who
fall under the XXX umbrella but are very much run as individual entities.
Question:
Would you agree that I’m going to need a SAN for every SIP domain’s access edge FQDN?
Thanks.

Remote panel and selective control access through Security with DSC

Hi Everyone,
I looked around to see what labview had to offer concerning security of remote panel and all the solutions I've found only propose full access to a remote panel, or none. The login page is not an option for me as everyone on the intranet can access the remote panel for monitoring but not everyone can have control to the buttons on it.
With the DSC module, I can give securities to the different controls and it works great with the exe, if I log in or out, the controls appears or disapears. Unfortunately, with remote panels, those controls are always visibles. Furthermore, when I log in from a remote panel, all the other remote panels get logged in with the same username and priviledges. When I log out, same thing, it logs out all the other remote panels. So the last guy who logs in gives its priviledges to every one else who is monitoring the remote panel at that time. when he logs out, he logs everyone out. I used a reentrant vi hoping that this would solve the problem but it didnt.
I would like to do what my colleague does with Advantech without any problem. He has only 1 .exe is running on the server and whoever connects to it through webserver, just needs to log in and he has all the priviledges of an guest, operator, admin, etc.. with access to controls and features accordingly. All this without interfering with the other people using the remote panels. This colleague has always been doubtful about the capabilities of labview to do SCADA systems and uptil now, I've been able to prove him wrong.. please help me continue
There are workarounds, using remote desktop to the server instead of webserver but it definitely is not as practical for the client and it will need quite some work to to synchronise all the exes open from the differents sessions, through the use of shared variables, binding and securities.
Thank you for your help.
Solved!
Go to Solution.

Thank you very much for your reply Jordan.
The NI security info is on the server (local domain with groups and users), and the running exes as well. The remote panel is not supposed to be accessed by internet, at least not yet, but just by the computers on the network and thinkline computers who are directly connected to the servers. the remote panel is accessed through the simple url: http://serverip/Application.html
There is no problem with accessing the remote panel of the exes on the server from another computer on the network. But there are several security issues when logging in and out with the NI Security Programmatic Login and logout VIs through the remote panel..
The 1st problem occurs when several people access the remote panel at the same time. In my setup, everyone is allowed to check what's happening on the front panel of the running exe, go through the tabs, check the graphs, the tables, etc.. , but only the administrators and the operators can send commands to the machines and the production line through this remote panel. Hence some buttons are accessible to all users, while others are only accessible depending on the privileges of the person logged in.
So like I said in the 1st post, I configured some buttons to be accessible only by the users of the admin group. When the exe runs, it's perfect, if I log in and out with an admin account, the buttons appears and disappear accordingly. But when I check the remote panel, those admin buttons are always visible, even if I am logged in as a guest or even logged out. Is it because the remote panel only needs minimum runtime engine and doesnt use the dsc runtime engine? if so, any work around?
Furthermore, another big problem is that if I login as an admin in one remote panel, then login as an guest in another remote panel, and then logout back from the first remote panel, it says: "User Domain/Guest logged out". Hence, my second login logged out my first user. I can actually see the admin buttons appearing and disappearing on the exe when login as admin and guest from the different remote panels.
So that's it, I would like my remote panel to behave like a normal scada system, with one exe running in the back (on the server) and with all the users accessing it through web server. Several users might/will access the remote panel at the same time and each of the users have a login/password that grant some of them the privileges to take some actions while giving the others only monitoring rights.
I hope that I have been more clear in this 2nd post,
Thank you again for your time.
Best Regards,
Tom.

ITunes library on Time Capsule accessed by multiple accounts - Not working!

I moved my iTunes library from an external hard drive to my new Time Capsule (it is in the Data folder). I can access it fine from my user account, but my husband who has an account on the same MacBook Pro cannot. When he tries to access the Data folder it has a red circle with a white line indicating that he does not have access. I have used Airport Utility to change the file sharing (enable file sharing with time capsule password and airport disk guests access read and write). But still he has no access on his account. Please help! Thanks!

I have my iTunes library on a Time Capsule. I can access it from more than one user account without the problems described here. In fact, I can access it from more than one account simultaneously if more than one machine is on. I have not tried accessing from multiple accounts using Fast User Switching.
The only think of that might be different is the location of the files. The iTunes Music folder is on the Time Capsule but not inside a Data folder. Each user's iTunes Library files are stored in the usual location in their home directory on the machines. iTunes' preferences are set to point to the folder on the Time Capsule.

Cluster in a box with Guest LDOMs : SCSI-3 reservations are not supported

Hello,
I recently installed the Oracle Solaris Cluster 4.1 in two simple Guest LDOMs (OVM for SPARC 3.1) with one T5140 server and SE 7110 as a shared storage. The some SE 7110 whole LUNs mapped as VDS devices to both LDOMs. After initial installation of Oracle Solaris Cluster 4.1 I observed issues with quorum device - both nodes cannot get SCSI3 (default) persistent reservation keys information at same time - only one node can be able to scrub. After switching cluster's global fencing option to "nofencing" (with "software" quorum implementation) both nodes can access quorum device successfully. So I can conclude that SCSI3 PGRs are not supported in this cases.
Unfortunately I didn't find any official and nonofficial documents info regarding this limitation. Could you please confirm or decline my suggestion.
Thanks.

I am answering on my question by itself
Planning the Oracle Solaris OS - Oracle Solaris Cluster Software Installation Guide
Fencing – Do not export a storage LUN to more than one guest domain on the same physical machine unless you also disable fencing for that device. Otherwise, if two different guest domains on the same machine both are visible to a device, the device will be fenced whenever one of the guest domains halts. The fencing of the device will panic any other guest domain that subsequently tries to access the device.
RTFM...

OS install in guest ldom fails on install-begin script

Hello,
I am installing solaris 10 u4 on guest ldom using DVD media loaded in the physical server. The guest ldom was able to boot from DVD media and goes through installation steps where I input the hostname, ip, etc upto the section where the install script prompts for "Remote Services Enabled". After this section, the installation drops to # prompt.
System identification complete.
/sbin/install-begin: test: argument expected
Exiting to shell.
When I do "cat /sbin/install-begin", this script shows incomplete. Earlier I checked the script in primary ldom under /cdrom/cdrom0/s1/sbin, and it looked ok and I am using the same DVD media which was used to build the base OS.
# cat /sbin/install-begin
#!/bin/sh
# Copyright 2006 Sun Microsystems, Inc. All rights reserved.
# Use is subject to license terms.
#ident "@(#)install-begin.sh 1.4 06/03/10 SMI"
# RESTRICTED RIGHTS LEGEND: Use, duplication, or disclosure by the Government
# is subject to restrictions as set forth in subparagraph (c)(1)(ii) of the
# Rights in Technical Data and Computer Software clause at DFARS 52.227-7013
# and in similar clauses in the FAR and NASA FAR Supplement.
# NOTE: At this point on non-Intel systems, the locale has
# not yet been configured
# NOTE: This script may be called from outside the window
# system (boot with "nowin" or kdmconfig/prtconf
# configuration failed) or within the window system
. /sbin/install-common
if install_debug_isset scripts ; then
===========================================
Not sure why the /sbin/install-begin script is not getting copied completly. Please let me if you have any input on this issue. I am planning to try with different DVD image to see if that helps.
Thanks for your help !!.

Thanks for your reply. Is there any fix for 6747687, I didnt see any in sunsolve. Here is the output from cdrw -Mv, the DVD I am using is the standard Sun DVD media.
# cdrw -Mv
Device : TSSTcorp CD/DVDW TS-T632A
Firmware : Rev. SR03 ()
Track No. |Type |Start address
1 |Data |0
Leadout |Data |1701760
Last session start address: 0
======================
Here is the df -k output from where it drops to shell prompt during install.
# df -k
Filesystem kbytes used avail capacity Mounted on
/virtual-devices@100/channel-devices@200/disk@3:b 174639 157917 0 100% /
/devices 0 0 0 0% /devices
ctfs 0 0 0 0% /system/contract
proc 0 0 0 0% /proc
mnttab 0 0 0 0% /etc/mnttab
swap 11498224 352 11497872 1% /etc/svc/volatile
objfs 0 0 0 0% /system/object
swap 11742576 244704 11497872 3% /tmp
/tmp/dev 11742576 244704 11497872 3% /dev
fd 0 0 0 0% /dev/fd
/devices/virtual-devices@100/channel-devices@200/disk@3:a 3210366 3210366 0 100% /cdrom
/cdrom/Solaris_10/Tools/Boot/usr 3210366 3210366 0 100% /usr
/platform/SUNW,SPARC-Enterprise-T5220/lib/libc_psr/libc_psr_hwcap1.so.1 174639 157917 0 100% /platform/sun4v/lib/libc_psr.so.1
/platform/SUNW,SPARC-Enterprise-T5220/lib/sparcv9/libc_psr/libc_psr_hwcap1.so.1 174639 157917 0 100% /platform/sun4v/lib/sparcv9/libc_psr.so.1
swap 11497880 8 11497872 1% /tmp/root/var/run

Guest Ldom bind error

I have created two guest ldoms ldg1 and ldg3 with the following config
Name: ldg1
State: active
Flags: transition
OS:
Util: 100%
Uptime: 23h 57m
Vcpu: 4
vid pid util strand
0 4 100% 100%
1 5 100% 100%
2 6 100% 100%
3 7 100% 100%
Memory: 2G
real-addr phys-addr size
0x8000000 0x88000000 2G
Vars: nvramrc=devalias disk /virtual-devices@100/channel-devices@200/disk@0
devalias vnet0 /virtual-devices@100/channel-devices@200/network@0
use-nvramrc?=true
Vdisk: vdisk0 vol1@primary-vds0
service: primary-vds0 @ primary
[LDC: 0x1]
Vnet: vnet0 [LDC: 0x4]
[Peer LDom: ldg3, mac-addr: 0:14:4f:f8:1:14]
mac-addr=0:14:4f:fb:fd:3
service: primary-vsw0 @ primary
[LDC: 0x2]
Vcons: [via LDC:3]
ldg1@primary-vcc0 [port:5000]
Vldcc: vldcc0 [Domain Services]
service: primary-vldc0 @ primary
[LDC: 0x0]
Name: ldg3
State: active
Flags: transition
OS:
Util: 100%
Uptime: 10s
Vcpu: 4
vid pid util strand
0 8 100% 100%
1 9 100% 100%
2 10 100% 100%
3 11 100% 100%
Mau: 1
mau cpuset (8, 9, 10, 11)
Memory: 2G
real-addr phys-addr size
0x8000000 0x108000000 2G
Vdisk: vdisk0 vol3@primary-vds0
service: primary-vds0 @ primary
[LDC: 0x0]
Vnet: vnet0 [LDC: 0x2]
[Peer LDom: ldg1, mac-addr: 0:14:4f:fb:fd:3]
mac-addr=0:14:4f:f8:1:14
service: primary-vsw0 @ primary
[LDC: 0x1]
Vcons: [via LDC:3]
ldg3@primary-vcc0 [port:5001]
Vldcc: vldcc0 [Domain Services]
service: primary-vldc0 @ primary
[LDC: 0x4]
When I create a third guest Ldom ldg2 , and issued the command
ldm bind ldg2
I receive this error
Unable to bind console client vcons0
After spending some time to resolve this issue i set the mau to 0 on ldg3
and unbound the domain and again executed a bind on ldg2 - it worked this time.
Has any one encountered this error earlier , I am aware of the fact that adding an mau to the control domain will prevent dynamic reconf of resources , but does adding an mau to guest ldom prevents you from binding other ldoms ? your thoughts are appreciated .

Hi, did you get the same error message about being unable to bind the console client?
The mau count has no bearing on the ability to bind other domains; only the ability to perform cpu DR on the domain that has the maus bound. The real determining factor is probably whether the other guest domain is bound or not. My hypothesis is that the LDom Manager is incorrectly concluding that there's a conflict for the console's TCP port, which is exactly the problem described in CR 6596652.
Can you also confirm that you specifically issued the ldm destroy & ldm create commands to completely destroy & recreate the domain that won't bind?
Thanks.
-Eric

ITunes Library on Time Capsule accessed by multiple accounts on same Mac

I moved my iTunes library from an external hard drive to my new Time Capsule (it is in the Data folder). I can access it fine from my user account, but my husband who has an account on the same MacBook Pro cannot. When he tries to access the Data folder it has a red circle with a white line indicating that he does not have access. I have used Airport Utility to change the file sharing (enable file sharing with time capsule password and airport disk guests access read and write). But still he has not access on his account. Please help! Thanks!

I have my iTunes library on a Time Capsule. I can access it from more than one user account without the problems described here. In fact, I can access it from more than one account simultaneously if more than one machine is on. I have not tried accessing from multiple accounts using Fast User Switching.
The only think of that might be different is the location of the files. The iTunes Music folder is on the Time Capsule but not inside a Data folder. Each user's iTunes Library files are stored in the usual location in their home directory on the machines. iTunes' preferences are set to point to the folder on the Time Capsule.

Guest LDOM disk access through multiple IO domains

Similar Messages

Maybe you are looking for