Ok, I have looked at these traces for a long while, and finaly I think I am seeing a pattern.
I've used this as an exercise to test out the latest version of WireShark (v0.99.5), which you can download from: http://www.wireshark.org/download.html (Some of the functionality may be in the final versions of Ethereal). WireShark loads & displays the snoop files with no problems. These are the menu options I found useful: "Analyse > Expert Info Composite" "Statistics > Service Response Time > SCSI..." ### first snoop file - "iscsinewfs_50GB.snoop": 23747 packets in 217 seconds. 4 previous segments lost 3 ACKed lost segments 871 Out-Of-Order segments 441 Fast retransmissions Many Duplicate ACKs, many Retransmissions, plus 100 'Window is full' 6 FINs and 6 RSTs 29 Scsi Writes(10), Service Response Time: Miniimum 0.40 seconds, Maximum 18.01 seconds, Average 5.11 seconds. ### second snoop file - "iscsinewfs_50GB-2.snoop"; 30413 packets in 404 seconds. 3 ACKed lost segments 1190 Out-Of-Order segments 743 Fast retransmissions Many Duplicate ACKs, many Retransmissions, plus 140 'Window is full' 6 FINs and 11 RSTs 405 Scsi Writes(10), Service Response Time: Miniimum 0.00145 seconds, Maximum 11.4 seconds, Average 0.78 seconds. In my experience, this is a high number of warnings. And the service response times look are a little high! ### Ok, let's look in detail at an example from the second snoop file. Packet 4619 @ 37.14s Initiator sends a SCSI: Write(10) followed by many '[Continuations to #4619]' packets, all of size 1514 bytes. Packet 4648 - 0.40s after previous packet. Packet 4659 - 0.80s after previous packet. Packet 4688 - 0.40s after previous packet. Packet 4700 - 0.80s after previous packet. Packet 4711 - 1.60s after previous packet. Packet 4724 - 3.20s after previous packet. Packet 4730 @ 44.40s Initiator still sending '[Continuations to #4619]'. Packet 4732 Initiator sends a 'NOP Out' (seeming to mean it has finished). Packet 4733 and 4734 Target sends an [ACK], followed by a [Dup ACK] Packet 4738 Initiator sends another '[Continuations to #4619]' but this is a retransmission. Packet 4739 & 4740 Target send an [ACK] and then send a iScsi 'Ready To Transfer' Packet 4741 Target sends a [FIN, ACK] Packet 4745 @ 44.81s Target sends a [RST], terminating the TCP stream. ### Ok, that is the pattern. If you work backwards from any of the targets [RST] packets, I think you will find a similar pattern to the above. What does this mean? I'm not sure. Someone with more expertise and experience is needed to give a definitive answer! Thanks Nigel Smith This message posted from opensolaris.org _______________________________________________ storage-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/storage-discuss
