HP StorageWorks Scalable File Share Release Notes - Version 2.3

Documentation updates and corrections 1–19
1.8 .1.6 Section 6.13 Step 13 : S c ann i n g MAC a d d resses
The scan mac command may not scale when applied to more than 20 servers at a time. HP recommends
proceeding by increments of 20 servers or less in such cases.
1.8 .1.7 Section 7. 2.3 iLO set t i ngs — iLO b rowser t a s ks
The same changes apply as for chapter "6.4.3 iLO settings — iLO browser tasks" above.
1.8 .1.8 Section 7.10 St e p 10 : S c an n ing M AC a d d re sse s
The scan mac command may not scale when applied to more than 20 servers at a time. HP recommends
proceeding by increments of 20 servers or less in such cases.
1.8.1.9 Appendix F SSSU script examples
Same remarks as for the section 5.2.4 above.
1. 8 . 2 HP StorageWorks Scalable File Share System User Guide
This section contains the following affected sections:
• Section 1.1 Product overview (Section 1.8.2.1)
• Section 5.7.1 Understanding quota tuning (Section 1.8.2.2)
• Section 6.2.10 Verifying email alert operation (Section 1.8.2.3)
• Section 7.9a Adding InfiniBand alias interfaces (Section 1.8.2.4)
• Section 7.11.1a Configuring multiple LNETS (Section 1.8.2.5)
• Section 7.11.2 Setting lustre.lnet (Section 1.8.2.6)
• Section 7.11.2a Setting lnet.ip2nets (Section 1.8.2.7)
• Section 9.32.1.1 Using the remote console (Section 1.8.2.8)
• Section 9.32.2 Troubleshooting iLO access (Section 1.8.2.9)
• Section 9.8 The scan mac command fails (Section 1.8.2.10)
• Section 9.9 Setting Lustre debug level (Section 1.8.2.11)
• Section 9.34 Troubleshooting failed SFS20 arrays (Section 1.8.2.12)
• Section B.10 Multiple IB interfaces bandwidth (Section 1.8.2.13)
1.8.2.1 Section 1.1 Product overview
Add a Section 1.1.4 Lustre MDS and OST services failover.
The Lustre MDS and OST services will fail over to the peer server in the following cases:
• The server hangs or goes down.
• Access to the underlying LUN failed.
• The network interconnect used for filesystem I/O failed, and that same interconnect is accessible from
the peer server. This allows failing over if the local interconnect interface fails, but not if the whole
interconnect is down (to avoid endless failover loops).