HP StorageWorks Scalable File Share Release Notes - Version 2.3

Problems fixed in HP SFS Version 2.3 1–15
• (7759) LustreError: Found inode with zero generation or link An inode generation of zero is perfectly
legal and Lustre should not complain about it.
• (7756) Client Enabler: When specifying the --prebuilt_ofed option on the command line, the
path specified by --ofed_src was not passed through the build_SFS_client.sh script
correctly.
• (7747) sfsconfig shouldn't need -s flag when the server is present and up.
• (7726) sfsconfig was incorrectly converting ldap entries to http.
• (7709) OSS quota related deadlock with multiple "lfs quota" commands.
• (7703) sfsconfig fixes required for support of InifiniBand memfree HCAS.
• (7682) Client panic in replay_lock_interpret; don't resend lock cancels and only resend if no
reconnect on ETIMEDOUT case.
• (7677) lfs quota block count statistics do not reflect real block usage /removing files did not always
reduce quota block count.
• (7664) Was logging /var/evlog.local on mds/admin when admin service was not running
there.
• (7635) Repeated "disk has been inserted" events were logged after a disk has been inserted.
• (7610) Collectl when enabled, could fill up /var with /var/log/collectl/system log files.
• (7608) start syslogd on OSS nodes, thereby stopping minilogd which should not have been
running on OSS nodes and slowly leaked user memory.
• (7579) Client crash in ll_imp_inval with Kernel BUG at filemap:485.
• (7492) sfsconfig: on failure to find common tcp subnet with server, find routed tcp network
instead.
• (7359) import reference counting in a log_commit thread was incorrect, leaving a window
open for the import to disappear while the thread was still using it, leading to a NULL pointer
dereference in ptlrpc_prep_req (ll_log_comt thread) and crashing an OSS node.
• (7277) Large file delete caused OSS nodes to be shot.
• (5988) configure server failed with error can't wait for variable "Done": would
wait forever.