Symptoms
[Networking mode]
Direct connection
[Operations]
1.Basic SAS services are configured. Snapshot, cache partition, SmartQoS, and LUN migration are configured.
2.Inject faults for a long period: reset controller 0B repeatedly, disconnect the front-end link abruptly, and disable LUNs' working controller.
After 0B is reset and restored, the link between the host and disk array is not restored.
Fault Diagnosis
1.Storage logs show that the host port that connects to storage port 0 is abnormal. The host sends an FC2 connection request but not sends an FC4 request. If the host does not send an FC4 request two seconds after the FC2 request, the storage sends an FC4 request. However, the host refuses this request and reports FCXLS_LS_RJT_INSUFFICIENT_RESOURCES (insufficient source).
2.The HBA on the Power5 host in this environment comes from another Power6 midrange computer.
The host information is as follows:
As described in IBM Power6:
http://www-01.ibm.com/support/knowledgecenter/POWER6/iphcd/fc5758.htm?cp=POWER6/2-3-12-2-3-28
FRU part number is 046K6838 or 003N5014.
As described in IBM Power5:
https://www-01.ibm.com/support/knowledgecenter/POWER5/iphcd_p5/fc5758.htm?cp=POWER5/1-10-3-2-11
FRU part number is 03N5014* or 03N5005**.
The normal output of this HBA on the Power5 is as follows:
HBAs on Power midrange servers are customized. HBAs have requirements for hardware environments. IBM said that third-party HBAs are not supported and HBAs cannot be replaced, and the company shall not be blamed for any problems caused thereby.
Solution
This problem is caused by incorrect hardware configurations. When this problem occurs, the HBA that connects to the 0B controller has been faulty, so you need to configure the HBA again. The following provides operation steps:
1.Runthe rmdev -l fcsX -R command to delete the configuration of fcsX. The following is an example of fcs0.
2.Runthe cfgmgr -l fcsX command to configure fcsX again. The following is an example of fcs0.
