Thursday Apr 28, 2016

Rolling Back the Job Log

When a job is scheduled to run routinely, it uses the same jobID each time. The Jobs pane will show that job running again and again. When you view the job's details, you're seeing the results of the most recent run of the job. But what about previous runs of that job? You can still view them, but you'll have to go back in time. If that's inconvenient, use the job's runID, which is the counter for each time the job runs.

Select the job and then click the View Details icon. In this example, the most recent time the job ran had the counter of Run-9. Click on the arrow next to the title of Job Details.

When a job runs routinely, the Run ID is drop-down list. This is where you can select a previous run.

Suppose we go back three runs to Run 6. Now the View Details icon will show the events in that run.

There's more about jobs and how to search for jobs in the documentation library.

Thursday Apr 21, 2016

Change of Address for the Enterprise Controller

To move the Enterprise Controller to a new server, you back it up on the original system and restore it on the new system. Each Proxy Controller has to be re-introduced to the Enterprise Controller at the new location. Before 12.3.1, the procedure was a little heavy-handed. Starting in 12.3.1, a new option for proxyadm streamlines the task. Let's line up the procedures:

Before 12.3.1 Version 12.3.1
As root, log in to the Proxy Controller. As root, log in to the Proxy Controller.

Stop the Proxy Controller:

# proxyadm stop -w

Stop the Proxy Controller:

# proxyadm stop -w

Unconfigure the Proxy Controller:

# proxyadm unconfigure -k

Update the connection.properties file:

proxyadm update -s <new_ip>

Reconfigure the Proxy Controller:

# proxyadm configure -s <newhostname>
            -u root -p <pwfilename>


Restart the Proxy Controller:

# proxyadm start -w

Restart the Proxy Controller:

# proxyadm start -w

Now, you need to know only the new IP address and you are only changing a value in a property file, not re-configuring the Proxy Controller. The entire workflow for moving controllers is in the Administration Guide in this topic: Backup and Restoring an Enterprise Controller.

Thursday Apr 14, 2016

What's In Your Datacenter

Starting in 12.3.1, Ops Center keeps count of your assets. Despite the name, the numbers are not a physical inventory of assets; they are the number of access points to the assets. An access point is the connection between the Enterprise Controller and an asset. Often, it's a 1-to-1 relationship, but it is common to have multiple connections between an asset and the Enterprise Controller. For example, a server will have an access point for its service processor and one for its operating system. A logical domain will have an access point for its control domain and one for its operating system. So what the asset counter really shows you is how many connections the Enterprise Controller is handling by type and how many connections each Proxy Controller is handling by type.

If you are only interested in a particular type of asset, use the Asset Counter tab in the Enterprise Controller's dashboard. Go to the Administration section of the navigation pane, then click the Asset Counter tab in the center pane. Let's say you are running a job that creates logical domains and you only need to check progress. You could always check the Jobs pane to see to see the job complete, but to see only each logical domain complete, refresh the Asset Counter tab and watch the count in the Ldoms column increase.

To investigate the access points, or if you wonder whether it's time to rebalance your Proxy Controllers, call the OCDoctor. Two of the OCDoctor's options now show the number of access points on the Enterprise Controller and on each Proxy Controller: the --troubleshoot option and the --collectlogs option. A new script in OCDoctor, AssetCount.sh, when you run it with its standard parameter, gives you the same information as you see in UI's Asset Counter tab.

# /var/opt/sun/xvm/OCDoctor/toolbox/AssetCount.sh standard

To drill down into the count, run the script with each of its other parameters: machine, agent, and all.

To see each Proxy Controller:

    #/var/opt/sun/xvm/OCDoctor/toolbox/AssetCount.sh machine

To see each Proxy Controller by type of access:

# /var/opt/sun/xvm/OCDoctor/toolbox/AssetCount.sh agent

To put the output together in one long listing,use the all parameter:

   # /var/opt/sun/xvm/OCDoctor/toolbox/AssetCount.sh all

The resulting output for a smallish datacenter, just 72 access points, looks like this: 


EC 72
Proxy Assets Zones Ldoms OVMGuests Servers Storages Switches ExadataCells MSeriesChassis MSeriesD
-------------------------------------------------------------------------------------------------
pc4   32     5     25    0         2       0        0        0            0              0
pc1   28     0     26    0         2       0        0        0            0              0
pc0   12     2     4     0         6       0        0        0            0              0

Proxy Agents Agentless SPs
--------------------------
pc4   25     2         0
pc1   1      1         0
pc0   5      5         5

Proxy 32 pc4
Zones 5 S11zone101 S11zone102 S11zone100 S11zone103 S11zone104
Ldoms 25 stdldom21 stdldom34 stdldom36 stdldom22 stdldom45 stdldom47 ...
OVMGuests 0
Servers 2 10.187.70.169 pc4
...
Proxy 28 pc1
Zones 0
Ldoms 26 stdldom21 stdldom34 stdldom36 stdldom22 stdldom45 stdldom47 ...
OVMGuests 0
Servers 2 10.187.70.171 pc1


By the way, if you're wondering at what point the number of assets affects performance, don't worry – Ops Center tells you if it's feeling the strain. When the Enterprise Controller exceeds 2700 access points, you'll get a Warning incident. At 3000 access points, you'll get a Critical incident. For a Proxy Controller, you'll get a Warning when it exceeds 450 access points and a Critical incident at 500 access points. So there's time and some headroom to make adjustments.

The Sizing and Performance Guide has more information, and if you want to adjust the number of access points each Proxy Controller is handling, see Rebalancing Assets.

Thursday Apr 07, 2016

What are JMX Credentials and What is Ops Center Doing With Them?

When you discover a Solaris Cluster, you're asked to provide ssh credentials and JMX credentials. You already know the ssh credentials but what about those JMX credentials? They're for the agent on the cluster's global node. The agent uses JMX so they're called JMX credentials. Think of them as agent credentials.

The only thing these credentials are doing is allowing the agent on the global node to respond to the Enterprise Controller. Without the JMX creds, you can discover and manage the cluster server itself, but nothing else. If you look in the log file, you'll see a message like "JMXMP provider exception java.net.ConnectException: Connection refused." With the JMX creds, Ops Center authenticates the agent, connects to the agent, and acquires all the agent's information about the global node.

JMX credentials can be anything convenient for you, like cluster1 and cluster2, and simple passwords. For all global nodes to use the same credentials, create one set in the discovery profile and run the discovery job. However, if for some reason, you need to use a unique set of credentials for each global node, create each set of credentials in a credential profile and then run a discovery job for each global node. You'll use the same discovery profile but change the credential profile for each job. You can still keep it simple, like cluster1node1 and cluster1node2.

For more information, take a look at the Oracle Solaris Cluster section of the Configuration Reference.

Tuesday Apr 05, 2016

Farewell and Hello

Folks,

I've been blogging about Ops Center for about eight years, off and on. It's been quite a ride, and I hope that over the years I've been able to help you, by answering questions and demonstrating features large and small.

However, I will be leaving Oracle soon. I'm handing the blog over to my coworker Barbara, who will be awesome. I'll let her introduce herself here, but I just wanted to say thank you for reading. Here's Barbara!

---

I always find it inspiring when someone sets a new course as Owen has. That's what I told him and now it comes to me to take a new role. I've been working on Ops Center for about 7 years, mostly in the storage and security corners of the product. I'm looking around for topics to spotlight, but I welcome your questions. If there's something new that you want me to explore or focus on, let me know.

Thursday Mar 31, 2016

Using Ops Center ASR For Unsupported Hardware

I saw a question recently about the limits of the Auto Service Request (ASR) feature in Ops Center:

"I have some hardware, including some switches, that are listed as supported in the ASR documentation but not in the Ops Center documentation. Can I still use Ops Center's ASR for them?"

In Ops Center, ASRs are generated through asset monitoring, so if a piece of hardware isn't supported in Ops Center, Ops Center can't generate ASRs for it. The Certified Systems Matrix talks a bit about how assets need to be listed as supported for both products for ASR through Ops Center to work for them.

The ASR documentation also has more information about what it supports. For products outside of Ops Center's umbrella, you'll need to use a separate ASR manager.

Thursday Mar 24, 2016

Searching in the Ops Center library

So, I've seen a few questions about the search function in the Ops Center doc library. This is a bit embarrassing, since the search function is supposed to answer questions instead of creating them. But, I figured I'd go over how the search works and how to tweak its settings.

Basically, the default setting for the search is based on where you are. If you do a search from the main page of docs.oracle.com,  it will cover literally every product library:


If you do a search from the Ops Center library, it will default to only searching the Ops Center library:

You can click Select in the categories on the left to see the current search options (and change them if need be):


If you're trying to find something within a specific book, then you can open up the book and then use the search box on the left:


Those are the basics of the search function in Ops Center.

Thursday Mar 17, 2016

Managing Guest Affinity and Disaffinity

When a guest is migrated to a new system, there are some cases where you want it to be placed with a specific guest, or want it not to be placed with a specific guest. Perhaps you have multiple guests that make up the parts of a three-tiered application, and you want to keep them together. Or perhaps you have two guests that are different nodes of an HA cluster, and putting them together would defeat the purpose.

Starting in Ops Center 12.3.1, you can use affinity tags to establish preferences for where guests go. You do this by selecting the guest, clicking Edit Tags, and using these tags:

  • affinityTags — This tag identifies the guest for other guests' affinities and disaffinities.

  • affinities — Which other guests the current guest should be placed with.

  • disaffinities — Which other guests the current guest should not be placed with.

For example, let's say that you have three groups of guests - A, B, and C - and you want to keep each group together. Here are the tags that you could assign to the guests in Group A:

 

  • affinityTags — Group A

  • affinities — Group A

  • disaffinities — Group B, Group C

There's more information about affinities and disaffinities in the Virtualization Reference document.

Thursday Mar 10, 2016

MAC Address Allocation for LDoms

MAC address collisions can cause a number of problems for logical domains. I've seen a few questions recently about how Ops Center allocates MAC addresses and how you can avoid or fix collisions, so I thought I'd address a few of them.

"How does Ops Center allocate MAC addresses?"

Ops Center uses the automatic range (00:14:4F:F8:00:00 - 00:14:4F:FB:FF:FF) to allocate MAC addresses to logical domains and network devices. You can use the manual range (00:14:4F:FC:00:00 - 00:14:4F:FF:FF:FF) for LDoms that you create manually.

"Is there a way to make sure that a MAC address won't have a collision before I bring it into Ops Center?"

The CLI can do this with the collisions mode. The list subcommand lists all known collisions, and the check subcommand checks to see if the specified IDs or addresses would have a collision.

"Does the UI keep track of all manually allocated MAC addresses?"

No, that information isn't collected in the UI.

Thursday Mar 03, 2016

Rebuilding an LDom Using Metadata

We recently got a question about rebuilding LDoms:

"Is there a way to use the metadata to rebuild an LDom on another CDom, if Ops Center was down or couldn't do Automatic Recovery?"

Yep. Beginning in 12.3.1, when you shut down and detach an LDom, the guest constraints are saved in an xml file in the /var/opt/sun/xvm/guests/ directory on the CDom. You can use this file to manually reconstruct the guest if necessary.

The xml file is named according to the guest name. For example, if you shut down and detach a guest named guest_1, the guest constraints are saved in the /var/opt/sun/xvm/guests/guest_1.xml file.

The Manage Logical Domains section of the docs discusses this, as well as a number of other options for managing LDoms.

Thursday Feb 25, 2016

Supported Remote Database Versions

I saw a couple of questions recently about database versions. One of the recent changes in Ops Center 12.3.1 was support for Oracle Database 12.1.0.2. If you have an embedded database, the database is automatically upgraded when you upgrade OC to 12.3.1.

The questions, though, were about what happens if you have a customer-managed database.

"If I'm upgrading to 12.3.1 with a customer-managed database, do I have to upgrade the DB to 12.1.0.2?"

 Nope. Oracle Database 11.2.0.3.0 or 11.2.0.4.0 are still supported for the remote DB, so you don't have to upgrade.

"What about a fresh installation?"

The same is true for a new installation. You can take a look at the Cert Matrix to see what's supported for what.

Thursday Feb 18, 2016

Support for Unified Archive (UAR)

Unified Archive, or UAR, is a new feature in Oracle Solaris 11.2. You can use UAR to take an image of an existing system, then use that image for provisioning. Folks have been asking about support for UAR in Ops Center since last year, so we made it a part of Ops Center 12.3.1.

You can import UAR images into Ops Center and provision operating systems, LDoms, and global zones from them, using Ops Center 12.3.1. This procedure explains how to do it, and goes over the prereqs.

Take a look at the What's New to learn more about the other new features in 12.3.1.

Thursday Feb 11, 2016

Viewing Assets in Maintenance Mode

In Ops Center, you can place assets into maintenance mode to temporarily stop alerts from being generated on them. This is useful if you're doing maintenance on some systems, hence the name; but particularly in a large environment you can lose track of what assets are in maintenance mode.

There are a couple of ways to tell. For a single asset, if you've selected it, the Actions panel will have a 'remove from maintenance mode' option if it's already in maintenance mode.

If you want to see all of the assets in maintenance mode, you can create a group that will automatically include them. Create a group with a single group rule like this:


This new group will then include all of your assets in maintenance mode.

Thursday Feb 04, 2016

Upgrade Issues

Now that some folks are upgrading to version 12.3.1, we've seen a couple of upgrade issues. There are a couple of issues that I've seen questions about which I can give you a solution for:

"When I try to upgrade to 12.3.1 on an S11 system, I get an error saying 'The configured Oracle Solaris 11 publisher, <publisher>, does not support the version of Oracle Solaris 11 that is installed on this system.' What do I do?"

The Oracle Solaris 11 publisher for your environment needs to include the full version of S11 that you're using. You can solve this problem by updating your repository.

"How much disk space do I need to have free for an upgrade?"

This will depend on your environment to some extent, but 100 GB is a good guideline.

Thursday Jan 28, 2016

CLI Enhancements in 12.3.1

One of the improvements in version 12.3.1 of Ops Center is a set of additional commands for the command-line interface.

The CLI lets you use many of the functions of Ops Center from the command line, and enables you to script some tasks. Beginning in 12.3.1, you can use the following subcommands in the CLI:

  • Plan: Lets you view and delete plans and credentials
  • Security: Lets you deploy and manage security certificates
  • Guest: Lets you view guests

There are also a number of new options for existing subcommands, giving you finer control. Take a look at the What's New and the CLI reference for more information.

About

This blog discusses issues encountered in Ops Center and highlights the ways in which the documentation can help you

Search

Archives
« July 2016
SunMonTueWedThuFriSat
     
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
      
Today