HP StorageWorks 4/16 HP StorageWorks Fabric OS 5.3.1a release notes (AA-RWEYJ- - Page 28

Linux kernel issue may occur on 2Gb switches running Fabric OS 5.3.0x caused by service daemon thread (fwd, nsd, fabricd) becoming unresponsive due to deadlock condition. This can trigger a panic, or snmpd hang. Fixed in Fabric OS 5.3.1 Fabric OS crash occasionally seen when Access Gateway enabled switch is rebooted with large number of Access Gateway devices in the fabric. Fixed in Fabric OS 5.3.1 During fastwrite, specific third party device drops a frame that is not 8 byte aligned. Fixed in Fabric OS 5.3.1 With 2Gb switches, hafailover under heavy load across ISLs, followed by portdisable/portenable of E-ports causes port blade fault with tx parity error (error 21). Fixed in Fabric OS 5.3.1 When PID-2 change occurs, Web Tools incorrectly translates the PID format in the NameServer and displays an incorrect account of the area/ports contained in a zone or alias member list. This can also happen when attempting to add to a zone or alias list, causing an incorrect physical port to be added. Fixed in Fabric OS 5.3.1 During stress test with 2Gb switch, when receive buffer overflow occurs, third party device port is stuck in offline state for more than 30 seconds. Portdisable/portenable will recover the port. Fixed in Fabric OS 5.3.1 Intermittent supportsave failure messages seen during supportsave loop. Fixed in Fabric OS 5.3.1 Traffic stops during FCIP testing with tape pipelining enabled, when host floods the switch with Name Server query. This occurs because the query response is not passed back to the host on the backbone switch through the FCR router. Fixed in Fabric OS 5.3.1 When executing fastwrite with tape pipelining with a slow responding tape device, the switch sent improper frame causing tape job to fail. Ultimately, this caused the initiator to lose access to the tape drive. Fixed in Fabric OS 5.3.1 When FICON host sends PLOGI directly to a device immediately after registering with the Management Server, PLOGI may be rejected Fixed in Fabric OS 5.3.1 because the routing tables have not been established. Incorrect GID FT response from Name Server caused host with multiple HBAs to log errors during HBA IOP reset. When host does its GID FT request, it gets a response for the other HBA though that HBA didn't register for FC4 type. Subsequent PLOGI, PRLI sequence to the non-FC4 type HBA is rejected and an error logged on the host. Fixed in Fabric OS 5.3.1 Automatic Fabric Configuration Server (FCS) policy distribution won't work properly on a future release with current Fabric OS 5.x in the Fixed in Fabric OS 5.3.1 fabric. Port scn removes logical paths from wrong port, resulting in some paths not coming online. Fixed in Fabric OS 5.3.1 If CALD restarts, it can no longer provide CAL cache data on EM, leading to possible SMI agent experiencing a reconnect issue. Fixed in Fabric OS 5.3.1 When Third party system sends PLOGI/PRLI frames with its DID as a broadcast format, the 2/128 forwards the frame as a broadcast frame to all switch ports, causing other devices in the fabric to malfunction and overextend the switch CPU. This issue only affects 2/128. Fixed in Fabric OS 5.3.1 Occasional kernel panic occurs during reboot. Fixed in Fabric OS 5.3.1 28