How to Perform 2 Basic Oracle 12c System Checks - dummies

How to Perform 2 Basic Oracle 12c System Checks

By Chris Ruel, Michael Wessler

You need to perform basic system checks to ensure the system is in a state that can support an Oracle 12c database. If the network is down, server is overloaded, or disk system has run out of space, your database may be impacted and display database errors, but the root problem is system related.

Investigate these key areas:

  • Network: Can you connect to the server or application?

  • Server utilization: What are the top processes on the server?

The Oracle 12c and the network

If you can’t connect to the database server, odds are good that your users can’t either. You have two easy ways to check this:

  • Ping to test server connectivity.

  • Log in to the server as the Oracle user.

From the DOS or Linux command prompt, type ping SERVER NAME to see whether the target server can be reached.

$ ping oralinux1
PING oralinux1 (192.168.2.121) 56(84) bytes of data.
64 bytes from oralinux1 (192.168.2.121): icmp_seq=1 ttl=64 time=0.020 ms
64 bytes from oralinux1 (192.168.2.121): icmp_seq=2 ttl=64 time=0.007 ms
64 bytes from oralinux1 (192.168.2.121): icmp_seq=3 ttl=64 time=0.007 ms
--- oralinux1 ping statistics ---
3 packets transmitted, 3 received, 0% packet loss, time 1999ms
rtt min/avg/max/mdev = 0.007/0.011/0.020/0.006 ms

In the preceding code, three ping packets were sent, and all three arrived successfully. Depending on the ping version, you get back slightly different output, but all outputs specify whether the server was reachable.

If the server comes back as unavailable, the problem is one of these things:

  • The server is shut down.

  • The network is down.

  • You’re prevented from pinging servers.

Sometimes security blocks the ping utility, so check whether it works before problems occur so you know the test is valid.

After you confirm the server can be reached, try actually logging in as the owner of the Oracle software if possible. Note that on some systems you have to log in as yourself (for security-auditing purposes) and then switch users to the Oracle software owner. This action confirms the server is not only running but also able to support a login attempt.

Oracle 12c and server utilization

If a program, process, or job is consuming all the resources on a server and has been doing so for a long enough time, a database can

  • Slow down

  • Be rendered unusable

  • Be killed (in rare cases)

The processes themselves may be one of three things: valid, a competing database (multiple databases can be on the same server), or an out-of-control, runaway process.

You must identify the programs that are running before you can determine whether they’re valid or hurting the system. There are several graphical tools to do this.

  • On Windows, use Windows Task Manager to see which applications are running (under the Applications tab). For more detail, use the Processes tab.

  • For Linux/UNIX, use the command top to display the top processes on a server and their process ID (PID). The output is text based and refreshed every few seconds.

    image0.jpg

image1.jpg

Additionally, at the top of the screen is the machine’s load average. This derived value reflects relative load on the server.

  • Values up to 3 are light and shouldn’t reflect performance problems.

  • Values in the teens reflect higher use of a busy system, and performance may suffer.

  • Values above 20 indicate a busy system where performance is likely impacted.

For Linux/UNIX, the uptime command helps you see system load:

$ uptime
 23:13:03 up 4 days, 10:27, 1 user, load average: 0.00, 0.00, 0.00

System load values are the same as with the top command:

  • The leftmost value is the current load.

  • The middle value is the load 5 minutes previous.

  • The rightmost value is the load 15 minutes prior.

Another useful value is the time since the last server restart. Obviously if users reported problems and you see the server recently rebooted a few minutes prior, the server reboot (or crash) is the likely culprit.

  • On UNIX Oracle Solaris systems, prstat is an alternative to top.

  • On HP-UX systems, the glance command is extremely useful.