Oracle Database Appliance
 

PLA-Support>Eng Systems>Exadata/ODA/SSC>Oracle Database Appliance>DB: ODA_EST
 

Applies to:

Symptoms

After replacing a failed Oracle Database Appliance (ODA) SSD drive in slot 22, we are now getting this as status for the replaced disk 22
  pd_19 /dev/sdae HDD ONLINE Good
  pd_20 /dev/sdap SSD ONLINE Good
  pd_21 /dev/sdar SSD ONLINE Good
  pd_22 /dev/sdax UNKNOWN PARTIAL PathsNotLoaded
  pd_23 /dev/sdaf SSD ONLINE Good

Changes

 SSD drive 22 was just replaced

Cause

You are encountering a known LSI controller bug. Unpublished Bug 13952653: LNX64-112-CMT: OAKD FOUND DISK TYPE UNKNOWN, EVEN IT FOUND SAS HDD SEAGATE.
This bug was talking about HDD Seagate SAS disks, but the same issue is possible for the SSD drives.
 
For example:
[root@oda1]# grep E1_S22 /etc/multipath.conf
  alias SSD_E1_S22_805731952
  alias UNKNOWN_E1_S22_805775895

[root@oda2]# grep E1_S22 /etc/multipath.conf
  alias SSD_E1_S22_805775895
 

Solution

In a first time, reboot the nodes and wait for 5-10 min after reboot before running any other action plan.

If this doesn't solve the issue, then perform the following steps.

from the above example,

{old id#} = 805731952
{new id#} = 805775895

1) remove the replacement SSD from slot 22.
2) From both ODA nodes, execute the following three commands:
 # multipath -r
 # multipath -F
 # multipath -v 3
3) Examine file /etc/multipath.conf on each of the ODA nodes. Remove both sections for SSD_E1_S22_{old id#} and UNKNOWN_E1_S22_{new id#} from /etc/multipath.conf file, if both lines are still there.
4) From both ODA Nodes, execute the following five commands:
  # oakcli stop oak
  # oakcli restart oak
  # multipath -r
  # multipath -F
  # multipath -v 2
5) Re-insert the replacement SSD into slot 22, and wait a few minutes.
6) At this point, an entry for "disk /dev/mapper/SSD_E1_S22_{new id#}p1 1 22 1" should have been added to /opt/oracle/extapi/asmappl.config. Please examine that file, and if the line is not present, please manually add it.
7) Next check the /etc/multipath.conf file, the /opt/oracle/extapi/asmappl.config file, and the output of “ls -al /dev/mapper/*S22*” from both nodes. Both Nodes should be in sync.
8) If everything is good so far, then check to see if that the replacement SSD 22 disk is added back to ASM. If it is not, then please manually add it, as follows:.
  +ASM> alter diskgroup redo add disk '/dev/mapper/SSD_E1_S22_{new id#}p1' name SSD_E1_S22_{new id#}p1;
 

References

This solution has no attachment