Basic Troubleshooting Techniques

The following are some of the most common configuration issues that prevent the Oracle ASM configuration from working properly.

If you encountering errors similar to:

• ORA-15075: disk(s) are not visible cluster-wide

• ORA-15032: not all alterations performed

This could be the result of improper settings for the I/O layer, meaning that your disks are not properly configured.

Perform the following tasks to verify that the settings are correct:

1. On the CipherTrust Manager, in the Host Group that was created for the RAC cluster, verify that the host group for this configuration does NOT have the Cluster Group option set ( this option is only for GPFS, which is not supported with CTE).

2. Ensure that the GuardPoints for the block devices are set at the Host Group level. This ensures that each node receives identical GuardPoints.

3. Verify that the GuardPoints are active on all nodes. When the GuardPoints are set, go to each node and verify that they are set and guarded, using the WebUI or the secfsd –status guard command. If they do not guard correctly, make sure the device names are the same across all nodes.

4. From ASM, make sure that the asm_diskstring parameter is modified to include the CTE devices and that the proper pathname is used, see Altering ASM_DISKSTRING on ASM.

Verifying Database Encryption

Option 1

The best way to verify the state of the data, without impacting anything in the existing environment, is to use the Oracle kfed command. You can run this command against the native path of the existing GuardPoints and make sure it returns with valid header information. If it returns valid information with the GuardPoint in place, then this confirms that the data is properly encrypted. If it returns with invalid header information, then that indicates that the data is either clear, double encrypted, or not in the expected encrypted state. The syntax for running this command would look similar to the following but will vary based on your environment.

/app/oracle/grid/product/19.0.0/grid/bin/kfed read /dev/<diskName>

If the location is properly encrypted, following is an example of the viewable output:

/app/oracle/grid/product/19.0.0/grid/bin/kfed read /dev/<diskName>

System Response:

kfbh.endian:                          1 ; 0x000: 0x01
kfbh.hard:                          242 ; 0x001: 0xf2
kfbh.type:                          124 ; 0x002: *** Unknown Enum ***
kfbh.datfmt:                         66 ; 0x003: 0x42
kfbh.block.blk:              1088904227 ; 0x004: blk=1088904227
kfbh.block.obj:              1558192170 ; 0x008: file=8234
kfbh.check:                  3321251423 ; 0x00c: 0xc5f6465f
kfbh.fcn.base:                932956641 ; 0x010: 0x379bc9e1
kfbh.fcn.wrap:               3040493590 ; 0x014: 0xb53a4016
kfbh.spare1:                 3806015223 ; 0x018: 0xe2db2ef7
kfbh.spare2:                 3794962182 ; 0x01c: 0xe2328706
60000000000D8000 01F27C42 40E75C23 5CE0202A C5F6465F
[..|B@.\#\. *..F_]
60000000000D8010 379BC9E1 B53A4016 E2DB2EF7 E2328706  [7....:@......2..]
60000000000D8020 CA2F30AD 522B4D21 99292639 004EBB34  [./0.R+M!.)&9.N.4]
60000000000D8030 A3896BE8 BD839D23 2204E19E 946C575C  [..k....#"....lW\]
60000000000D8040 4CE2218F 35E1B101 AF751A70 780E6D6E  [L.!.5....u.px.mn]
60000000000D8050 5E7E6A38 C600ED5F 929047C4 DF372A8E  [^~j8..._..G..7*.]
60000000000D8060 E103152D BA87CC03 11A7D963 9D72FCE1
[...-.......c.r..]
60000000000D8070 1EC6B48B 03EE869F 61D651F9 E7614957  [........a.Q..aIW]
60000000000D8080 810E0353 9C461F49 69569733 501D19EF  [...S.F.IiV.3P...]
60000000000D8090 B268002B 4F9457B6 BDB04AC5 D3D07446  [.h.+O.W...J...tF]
60000000000D80A0 FD9EE5E0 9B46CB66 30D10B22 F63AB77E  [.....F.f0..".:.~]
60000000000D80B0 6FF79075 4BBD1FAD 8F226188 7774300D  [o..uK...."a.wt0.]
60000000000D80C0 A809B6FB E1F1C80B B5760E68 9747D97D  [.........v.h.G.}]
KFED-00322: Invalid content encountered during block traversal: [kfbtTraverseBlock][Invalid OSM block type][][124]

Option 2

The second option to verify the state of the data is to use the dd command. This requires you to specify some blocks and write it out to a file. After this completes, read the file using the strings command. You can do this while the device is in use. In the example below some sectors are skipped and it only dumps 10000 counts.

For example:

dd if=/dev/asm_data2p1 of=/tmp/dd2.out skip=1047 count=10000

Option 3

The third option to verify the state of the data without impacting the existing environment is to use the strings command.

Note

The strings command cannot read a busy or large device.

You can run this command against the native path (/dev/<deviceName>) of the existing GuardPoints (/dev/secvm/dev/<deviceName>). By executing the strings command against the native path strings /dev/devicename | more, this does not go through the SecVM device and therefore is not be decrypted. If it is encrypted the output should contain illegible text.