Quantum| Fiber Connectivity |
This page provides information on: answers these questions:
To find the connectivity status of each port in an IOB, open its supportDump file with WordPad and look for the phrase: "fcShow 2".
Several sections appear under "fcShow 2". The first section, "Fibre Channel Controllers", shows the current connectivity status for ports 1 - 6. In a standard configuration, ports 1 and 2 are configured for host/SAN attachment, and ports 3 - 6 are configured for attachment to tape drives.
************ fcShow 2 ************
---------------------------------------------------------------------------------------------
Fibre Channel Controllers
---------------------------------------------------------------------------------------------
Ctlr : PCI Addr : ISP : Firmware : Firmware : Loop : Fabric : Port : Ext. : Link
Id : Bs Dv Fn : Type : State : Version : ID : Attached : Mode : FIFO : Speed
---------------------------------------------------------------------------------------------
1 : 00 01 00 : 2312 : Ready : 3.03.19 : PtoP : Yes : Targ : None : 2 Gbps
2 : 00 01 01 : 2312 : Ready : 3.03.19 : PtoP : Yes : Targ : None : 2 Gbps
3 : 00 02 00 : 2312 : Ready : 3.03.19 : 0 : No : Init : None : 2 Gbps
4 : 00 02 01 : 2312 : Ready : 3.03.19 : 0 : No : Init : None : 2 Gbps
5 : 00 03 00 : 2312 : Ready : 3.03.19 : 0 : No : Init : None : 2 Gbps
6 : 00 03 01 : 2312 : Ready : 3.03.19 : 0 : No : Init : None : 2 Gbps
---------------------------------------------------------------------------------------------
Important field values include the following (some fields are not covered below, because they don't provide information that is useful for field service):
| Ctlr ID | FC port on the blade, with port 1 being the bottom and port 6 the top. |
| Loop ID | PtoP if the port is linked in Point to Point mode; [a number] if the port is linked in Loop mode; or None if the port is not linked. |
| Fabric Attached | Yes if the library did an FLOGI (Fabric Login) to the switch. Generally, if the Loop ID is PtoP, it will be Fabric Attached. |
| Port Mode | Shows if the port is set to be a target (SAN or host attachment) or Initiator (Device attachment). |
| Ext. FIFO | This field doesn't provide useful information. |
| Link Speed | The speed at which the port is currently linked, or 0 if not attached. |
Here is an example of a 7404 blade with port 2 not connected to anything:
---------------------------------------------------------------------------------------------
Fibre Channel Controllers
---------------------------------------------------------------------------------------------
Ctlr : PCI Addr : Device : Firmware : Loop : Fabric : Port : Link
Id : Bs Dv Fn : Type : Version : ID : Attached : Mode : Speed
---------------------------------------------------------------------------------------------
1 : 00 02 00 : LSIFC949X : 1.03.16 : PtoP : Yes : Targ : 4 Gbps
2 : 00 02 01 : LSIFC949X : 1.03.16 : None : No : Targ : 0 Gbps
3 : 32 02 00 : LSIFC949X : 1.03.16 : 0 : No : Init : 4 Gbps
4 : 32 02 01 : LSIFC949X : 1.03.16 : 0 : No : Init : 4 Gbps
5 : 32 03 00 : LSIFC949X : 1.03.16 : 0 : No : Init : 4 Gbps
6 : 32 03 01 : LSIFC949X : 1.03.16 : 0 : No : Init : 4 Gbps
To find the error statistics for each port (SFP) in an IOB, open its supportDump file with WordPad and look for the phrase: "fcShow 2".
There are several sections found under "fcShow 2". Earlier, we looked at the first section for current link status. The next section has breakouts for each port, entitled "Fibre Channel Controller X", where X is the port number, 1 - 6. We will talk about that section later.
The third section again contains breakouts for each port. These breakouts are named the same as in the previous breakout section, but the content is different. This is the information we are looking for there. Three examples are shown:
A target (host) port on a 7404 blade, port 1:
--------------------------------------------------------------
Fibre Channel Controller 1
--------------------------------------------------------------
Loop ID: None, AL_PA: None
Port Connection Type = 1
Port Mode = 1
FCAL Position Map (Relative to Loop Master)
Pos AL_PA
-----+-------
Link Stats for Loop ID 255 on FC1
Link Failures -- 01
Loss of Sync -- 026
Loss of Signal -- 026
Prim. Seq. Prot. Err -- 00
Invalid Transmission -- 00
Invalid CRC -- 00
A target (host) port on a 6404 Blade:
--------------------------------------------------------------
Fibre Channel Controller 1
--------------------------------------------------------------
Port ID: 0x10000
Link Stats for FC1
Link Failures -- 0
Loss of Sync -- 1
Loss of Signal -- 1
Prim. Seq. Prot. Err -- 0
Invalid Transmission -- 0
Invalid CRC -- 0
An initiator (tape drive) port on a 6404:
--------------------------------------------------------------
Fibre Channel Controller 4
--------------------------------------------------------------
Loop ID: 0, AL_PA: 0xef
Port Connection Type = 0
Port Mode = 18
FCAL Position Map (Relative to Loop Master)
Pos AL_PA
-----+-------
000 : 0x26 (loop master)
001 : 0xef
Link Stats for Loop ID 0 on FC4
Link Failures -- 0
Loss of Sync -- 5
Loss of Signal -- 5
Prim. Seq. Prot. Err -- 0
Invalid Transmission -- 0
Invalid CRC -- 0
Important Entries Under "Link Stats"
Not all entries are important to discuss at this point. The significant ones are covered below.
Link Failures, Loss of Sync, and Loss of Signal: These should not be high numbers (in the hundreds would be high, for example). Switch problems or physical layer problems can cause these numbers to increase.
There is no certain number that is acceptable or too high -- this varies. In the example above, the 5 for loss of sync and loss of signals could have several causes, such as someone unplugging cables to the IOB or rebooting a switch, or even other events from some time ago.
It is important to check a given snapshot against another one taken later, to see how many new events have occurred during that time period. A well-running ecosystem should not have new unexplained events in these three catagories.
Prim. Seq. Prot. Err: These numbers should be low or zero. If not, escalate to the RTS.
Invalid Transmission: These errors can be recoverable and can get high (up to 100) on some systems, especially on initiator (drive-connected) ports, without causing problems or interrupting I/O to the device. However, numbers in the hundreds of thousands always indicate a problem and are usually accompanied by Link Failures, Loss of Sync, or Loss of Signals, although the numbers for those will be much lower.
The cause for these errors is usually a bad SFP or connection in the drive sled, a bad cable between the drive and the blade, or a bad SFP in the blade. Please note that:
NOTE: After each change is made to try to eliminate high Invalid CRC rates, take two snapshots, one immediately after the change, and one some time later. The act of replacing or reseating a fiber cable or SFP will generate Invalid CRC's and other Link errors, so you need to use a snapshot taken immediately after the remedial action was performed in order to get a baseline for comparison against future snapshot link statistics.
Invalid CRC's: These are usually fatal I/O errors. If found between a drive and blade, suspect the drive and blade SFP's, and the FC cable between them. You are more likely to see these on the host ports (1 and 2). They indicate a problem in the ecosystem or SAN between the host HBA and the IOB, including both endpoints and all hardware in between. Ongoing problems with CRC errors should be escalated to the RTS if no faulty hardware is found in the SAN.
NOTE: Some of the link events logged in the Link Stats section of the supportDump are also logged as "events" in the loggerDump section of the supportDump. This will be covered in a later section of this material.
This page was generated by the BrainKeeper Enterprise Wiki, © 2018 |