EMC AX100 Release Notes - Page 14

Fixed problems Problem AX100SC, AX100SCi, AX100, AX100i - Under certain situations, the write cache cannot be enabled after an upgrade. This may also cause the SP to panic. (124595) AX100SC, AX100SCi, AX100, AX100i - If a faulty HBA attempts to log in to an SP using an illegal (all zeros) WWN, both SPs will panic simultaneously. This may result in a data unavailable situation. (124275) AX100SCi, AX100i - The SCSI Mode Sense test component of the Microsoft Hardware Compatibility Test suite for certification fails intermittently. (126406) AX100SCi, AX100i - When the iSCSI Internet Storage Naming Service (iSNS) is active, an upgrade may fail due to excessive processor use. (123009) AX100SC, AX100SCi, AX100, AX100i - CLARiiON storage systems connected to Cisco switches may report hosts as logged in even when those hosts have been disconnected or removed from the zone. When the hosts are properly connected to the network/storage system, all I/O requests are handled normally. (118562, 116887) AX100SC, AX100SCi, AX100, AX100i - Storage system may not log out initiators properly if all initiators are removed from the zone. (118871, 118872) Details Fixed in version: 02.19.100.5.005 Solution: A sanity check failed to catch the fact that there were virtual disks with dirty cache pages and permitted changes to an internal data structure containing a list of cache dirty virtual disks. This situation left the list in an indeterminate state, which ultimately led to the problem. The sanity check was modified to properly check for cache dirty virtual disks prior to upgrade to allow the system to take the appropriate action. Fixed in version: 02.19.100.5.005 Solution: The invalid WWN happens to be one that the SCSI target driver uses internally. The target driver believes that the host is already logged in and subsequently panics the SP to clear the situation. The software has been modified to reject the login of invalid WWNs without a panic and return an appropriate error message with the reason for the rejection. Fixed in version: 02.19.100.5.005 Solution: A buffer containing the information representing the mode page was not being updated with new data. Subsequently, there were times during the test where the data was appropriate and the test passed and other times when it was not refreshed and the test failed. The SCSI HCT test passes now that the buffer data is always updated. Fixed in version: 02.19.100.5.005 Solution: Instantiating a particular object repeatedly causes the system to produce inaccurate processor use statistics. The iSNS executable has been fixed to not instantiate this object repeatedly. Fixed in version: 02.19.100.5.005 Solution: Cisco switches can generate host source IDs that are non-standard. These nonstandard values resulted in the hosts being incorrectly classified as being on a local loop and therefore excluded from zone checking. This then resulted in the hosts not being properly logged out after removal from the zone. Fixed the code to verify that the local link is running as a loop before excluding from zone checking. Fixed in version: 02.19.100.5.005 Solution: When the storage-system port is the only port in the zone, the code failed to reset the count of initiators in the zone to 0, using the previous zoning information to log out initiators. Corrected the code to properly set the count to 0 in this case so initiators are logged out properly. 14 EMC AX100-Series Integrated Management Software Release Notes