On Apr 1, 2009, at 6:59 AM, James Smart wrote:
The largest issue I have is - what attributes are really fc/fcoe
specific ?
DCBX and PFC are arguably NIC-related parameters and have no
business being
under the fc transport. Additionally, whatever we pick, we had
better put
the same or like parameters with lib_fcoe-supporting adapters in the
same
place.
This is very muddy as some adapters will want present a fc/scsi
function
only, hiding the nic completely; others may present a nic function and
an fcoe function, and physically share the nic; while others will have
only the nic and a bunch of software, or a nic with super-features for
fcoe. What object belongs where for what attribute ?
I agree. With various ways the adapters can support the FCoE
functionalities, it's big issue to identify FC and FCoE specific
attributes.
I felt there should be separate statistics for those adapters where
both FCoE and NIC functions are exposed, but use only the FCoE
function and FC transport layer similar to other FC adapters. Some of
the existing statistics does not give complete picture for such
adapters and hence needed more additions. Though adding few of the MAC
statistics gives better picture, but are totally non-FC specific and a
question whether to be supported in FC transport layer? But without
NIC in this case, it needs some place holder and I thought statistics
under FCoE may be a better place.
Another thing that should be brought up is the presentation model when
there are multiple FCF's that an FCOE adapter can talk to. I'm a fan
of
having a new fc_host for every new *initiator* context on a fabric.
Meaning, there's a fc_host for each N_Port_Id on each fabric (which is
what we have been doing for NPIV and VSANs). Mean an FCOE port, which
sees multiple FCFs, or contacts the same FCF on different vlans (which
map to different VSANs) need to be separate fc_hosts. Additionally,
we
should also consider a bit, what and how do we manage when there are
multiple FCF's into the same fabric. (Note: this pushes again on why
isn't FC a "bus" rather than the top thing, usually a pci function,
must
be a scsi_host ?).
A few more comments inline...
I agree for having a separate host for each N_Port_Id. This gives
better distinction and flexibility with multiple FCF support.
-- james s
Giridhar Malavali wrote:
Hi SCSI mailing list,
I'd like to propose following additions to sysfs to export
statistics
of FCoE host bus adapters.
The additions can be broadly divided into capabilities of FCoE HBA
adapter and its statistics.
I am thinking of extending HBA specific informations inside the
fc_host(/sys/class/fc_host/hostX) and make a seperate
attribute_group(/sys/class/fc_host/hostx/fcoe_statistics/) for fcoe
statistics.
I disagree with the "statistics" in the name unless it truly is a
statistic. I do agree with, for fcoe-specific fc information, a
subdirectory (or attribute group) that is under the fc_host.
(such as /sys/class/fc_host/hostX/fcoe and if there are fcoe-specific
statistics, in a directory /sys/class/fc_host/hostX/fcoe/statistics).
The various different FC link errors (LESB) in the fc_host/statistics
can be mapped to counter parts in MAC. I saw a proposal(09-204v0) for
addition of these statistics to FC-BB-5 in T11, by Roger Hathorn. I
think those MAC statistics can be used as FCoE specific statistics.
FCoE HBA specific information
1) enode_mac_address /* Factory programmed MAC address */
2) vn_port_mac_address; /* Current programmed MAC address */
3) fcf_mac_address; /* FCF mac address */
4) vlan_id: /* Local VLAN ID */
5) mac_addressing_model /* Whether SPMA or FPMA */
These make a lot of sense to go into the .../hostX/fcoe directory.
As we
look closer, we'll probably extend this list.
Current DCBX parameter details:
PGID (Priority group ID)
1) pgid_priority_group_0; /* Priority group ID of priority
group
0 */
2) pgid_priority_group_1; /* Priority group ID of priority
group
1 */
3) pgid_priority_group_2; /* Priority group ID of priority
group
2 */
4) pgid_priority_group_3; /* Priority group ID of priority
group
3 */
5) pgid_priority_group_4; /* Priority group ID of priority
group
4 */
6) pgid_priority_group_5; /* Priority group ID of priority
group
5 */
7) pgid_priority_group_6; /* Priority group ID of priority
group
6 */
8) pgid_priority_group_7; /* Priority group ID of priority
group
7 */
Bandwidth assignment per priority group
1) priority_group_0_bw_percentage; /* Priority group 0 bandwidth
percentage */
2) priority_group_1_bw_percentage; /* Priority group 1 bandwidth
percentage */
3) priority_group_2_bw_percentage; /* Priority group 2 bandwidth
percentage */
4) priority_group_3_bw_percentage; /* Priority group 3 bandwidth
percentage */
5) priority_group_4_bw_percentage; /* Priority group 4 bandwidth
percentage */
6) priority_group_5_bw_percentage; /* Priority group 5 bandwidth
percentage */
7) priority_group_6_bw_percentage; /* Priority group 6 bandwidth
percentage */
8) priority_group_7_bw_percentage; /* Priority group 7 bandwidth
percentage */
Priority based flow control (PFC)
1) pfc_enabled_priority_0 /* Flow control is enabled in both
direction for priority 0 */
2) pfc_enabled_priority_2 /* Flow control is enabled in both
direction for priority 1 */
3) pfc_enabled_priority_3 /* Flow control is enabled in both
direction for priority 2 */
4) pfc_enabled_priority_4 /* Flow control is enabled in both
direction for priority 3 */
5) pfc_enabled_priority_5 /* Flow control is enabled in both
direction for priority 4 */
6) pfc_enabled_priority_6 /* Flow control is enabled in both
direction for priority 5 */
7) pfc_enabled_priority_7 /* Flow control is enabled in both
direction for priority 6 */
8) pfc_enabled_priority_8 /* Flow control is enabled in both
direction for priority 7 */
As mentioned, I have heartburn with this under the fc_host.
But if there is no NIC - where should it go ?
Statistics:
1) fcoe_tx_frames; /* number of FCoE transmit
frames */
2) fcoe_tx_words; /* number of tx words */
3) fcoe_rx_frames; /* number of FCoE receive
frames */
4) fcoe_rx_words; /* number of rx words */
5) fcoe_rx_drop_frames; /* number of FCoE dropped
receive frames */
What's the difference between these and the normal FC statistics ?
They are the same. I though of keeping it for comparison reasons. We
can keep it in case of separate fcoe group in fc_host or can be removed.
6) fcoe_tx_pause_pkts; /* number of FCoE transmit
PAUSE packets */
7) fcoe_rx_pause_pkts; /* number of FCoE receive
PAUSE packets */
8) fcoe_tx_cbfc_pause_frames_0; /* number of class based flow
control transmit PAUSE frames on PG 0 */
9) fcoe_tx_cbfc_pause_frames_1; /* number of class based flow
control transmit PAUSE frames on PG 1 */
10) fcoe_tx_cbfc_pause_frames_2; /* number of class based flow
control transmit PAUSE frames on PG 2 */
11) fcoe_tx_cbfc_pause_frames_3; /* number of class based flow
control transmit PAUSE frames on PG 3 */
12) fcoe_tx_cbfc_pause_frames_4; /* number of class based flow
control transmit PAUSE frames on PG 4 */
13) fcoe_tx_cbfc_pause_frames_5; /* number of class based flow
control transmit PAUSE frames on PG 5 */
14) fcoe_tx_cbfc_pause_frames_6; /* number of class based flow
control transmit PAUSE frames on PG 6 */
15) fcoe_tx_cbfc_pause_frames_7; /* number of class based flow
control transmit PAUSE frames on PG 7 */
16) fcoe_rx_cbfc_pause_frames_0; /* number of class based flow
control recieve PAUSE frames on PG 0 */
17) fcoe_rx_cbfc_pause_frames_1; /* number of class based flow
control recieve PAUSE frames on PG 1 */
18) fcoe_rx_cbfc_pause_frames_2; /* number of class based flow
control recieve PAUSE frames on PG 2 */
19) fcoe_rx_cbfc_pause_frames_3; /* number of class based flow
control recieve PAUSE frames on PG 3 */
20) fcoe_rx_cbfc_pause_frames_4; /* number of class based flow
control recieve PAUSE frames on PG 4 */
21) fcoe_rx_cbfc_pause_frames_5; /* number of class based flow
control recieve PAUSE frames on PG 5 */
22) fcoe_rx_cbfc_pause_frames_6; /* number of class based flow
control recieve PAUSE frames on PG 6 */
23) fcoe_rx_cbfc_pause_frames_7; /* number of class based flow
control recieve PAUSE frames on PG 7 */
24) fcoe_tx_priority_pkts_0; /* number of priority 0
based transmit packets */
25) fcoe_tx_priority_pkts_1; /* number of priority 1
based transmit packets */
26) fcoe_tx_priority_pkts_2; /* number of priority 2
based transmit packets */
27) fcoe_tx_priority_pkts_3; /* number of priority 3
based transmit packets */
28) fcoe_tx_priority_pkts_4; /* number of priority 4
based transmit packets */
29) fcoe_tx_priority_pkts_5; /* number of priority 5
based
transmit packets */
30) fcoe_tx_priority_pkts_6; /* number of priority 6
based transmit packets */
31) fcoe_tx_priority_pkts_7; /* number of priority 7
based transmit packets */
32) fcoe_rx_priority_pkts_0; /* number of priority 0
based received packets */
33) fcoe_rx_priority_pkts_1; /* number of priority 1
based received packets */
34) fcoe_rx_priority_pkts_2; /* number of priority 2
based received packets */
35) fcoe_rx_priority_pkts_3; /* number of priority 3
based received packets */
36) fcoe_rx_priority_pkts_4; /* number of priority 4
based received packets */
37) fcoe_rx_priority_pkts_5; /* number of priority 5
based received packets */
38) fcoe_rx_priority_pkts_6; /* number of priority 6
based received packets */
39) fcoe_rx_priority_pkts_7; /* number of priority 7
based received packets */
Aren't these really NIC-level statistics too ? what makes them so
fcoe-ish ?
Yes, they are NIC statistics, but once again I thought of having these
for better clarity from FCoE protocol perspective.
Since there is no specific place holder for this, I thought of adding
this in FC transport.
Thanks,
Giridhar Malavali
--
To unsubscribe from this list: send the line "unsubscribe linux-
scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
