Hi, In reference to "Sun Open Storage" aka s7000 series and OpenSolaris.
CR6770534 was corrected long ago and is present since snv_107 and s10u8_03. The bug applied the the iscsitgtd service which is not current. COMSTAR replaces it now. Yes, currently ZIL is used for iSCSI writes provided that the customer did not change the wcd=true option to wcd=false which allows the application to set it's own SCSI CDB write flag (as you are aware of now). Large writes (tunable) bypass a SLOG vdev and go directly to the disk vdevs which are still part of a ZIL transaction group and are only returned committed on a ZIL transaction complete. So there is NO data loss from a storage subsystem perspective that can occur. Sync writes on any system do not result in 0 data loss, it simply means the data write is committed to stable storage as per the calling request. And the application must decide if it can continue based on that low level operation. Keep in mind that iSCSI etc. is still subject to typical data loss on the network adapter ring buffers as well, which is applicable to every iSCSI based storage system. Regards, Mike http://blog.laspina.ca/ -- This message posted from opensolaris.org _______________________________________________ storage-discuss mailing list storage-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/storage-discuss
