This article is translated from the blog article posted by EMC Enterprise Storage Division (USD) Remote Support, China team.

Host not logging into the storage is a common issue. It usually occurs after the implementation of a new device or a host reboot. When it happens, the host will show "Registered: Yes; Logged In: No" in Unisphere:

    

Some paths from host to storage will be unavailable, which may affect the performance and availability. After the initial investigation, host or storage administrator could not find any changes which can be the cause of the problem. As the transportation infrastructure of the SAN network, FC switch is a key part that needs to be carefully checked.

First let’s talk about the communication process of two devices, using the flowchart in Brocade’s BCFA training guides:

    

Note: Generally speaking, the port type of the target devices (host HBA ports, storage front-end ports) is the N_Port, the port at the switch side is the F_Port.

As we can see, there are five phases in the initialization process.

Step 1.Initialize the link

After the device connects to the switch, first it needs to initialize the link, sync-up the signal, then negotiate the port speed. In this phrase, we should make sure that the intensity is strong enough so that the switch can receive the normal signal. You can use the following commands to check it on the switch:

Brocade Switch

sfpshow[slot/]port -f

Note: -f parameter requires FOS 6.3 or higher version.

Cisco Switch

show interface fc x/y transceiver details

Replace the red part with your slot/port numbers.

    

As you can see, RX Power is the power in dBm of the received signal. EMC recommends it should not be less than -7 dBm. Otherwise you have to check the fiber cables, patch panels and end devices.

Then we should verify if the device speed matches the switch ports speed. Generally for SAN switches, the auto-negotiation speed of the SFP module or port is the maximum and two lower speeds. For example, a 16G switch’s SFP will auto-negotiate at 4/8/16Gbps speed. If you manually specify the speed, you should make sure the end device can work at one of these three speeds. You can use the following commands to check the speed of the SFP module:

Brocade Switch

    

Cisco Switch

    

If the above two items are Ok, the link should be up.

Step 2.FLOGI

Next, the end device will initiate the FLOGI (Fabric Login) process. It is similar to the DHCP in IP network. The device sends the request to Fabric Login Server (address 0xFFFFFE), the switch replies with a 24bit FC address (just like an IP address) for routing.

    

We can use the following commands to verify:

Brocade Switch

portloginshow[slot/]port

Example:

DS-5100B:admin> portloginshow 5

Type  PID     World Wide Name        credit df_sz cos

=====================================================

fe  01050050:06:01:6c:3c:e0:32:e78  2048   8 scr=0x3

ff010500 50:06:01:6c:3c:e0:32:e7     4   128 8  d_id=FFFFFC

Cisco Switch

showflogi database interface fc x/y details

Example:

`show flogi database details`

--------------------------------------------------------------------------------------------

INTERFACE        VSAN    FCID           PORT NAME               NODE NAME            FLAGS

--------------------------------------------------------------------------------------------

fc2/20           700   0x290140  50:00:09:74:08:31:12:3450:00:09:74:08:31:56:78 P

If there is a PWWN and FC address, then the FLOGI process is successful.

Step 3.PLOGI to Name Server

Next, the end device will initiate the PLOGI (Port Login) process. The Switch will write the information of the device to Name Server (a service on switch, address 0xFFFFFC). This information includes PWWN, NWWN, FC4 type and so forth.

Then the end device will initiate SCR (State Change Registration) to the Fabric Controller (another service on switch, address 0xFFFFFD), to apply for receiving the RSCN (Registered State Change Notification).

Every time the FC topology changes or the device becomes up or down, the switch will send the notification to the devices that finish the SCR.

Use the following commands to verify if the PLOGI process to Name Server is successful:

Brocade Switch

nsshow-r

Example:

Type Pid    COS     PortName                NodeName                 SCR

N    030000; 3;50:00:09:74:08:35:12:34;50:00:09:74:08:35:56:78; 0x00000003

FC4s: FCP

PortSymb: [94] "SYMMETRIX::000292601357::RAF- 7fA::FC::5875_286+::EMUL F80F0000 381AA71D 802DB0 07.02.13 14:26"

NodeSymb: [38] "SYMMETRIX::000292601357::FC::5875_286+"

Fabric Port Name: 20:00:00:05:33:7d:90:ab

Permanent Port Name: 50:00:09:74:08:35:12:34

Port Index: 0

Share Area: No

Device Shared in Other AD: No

Redirect: No

Partial: No

LSAN: No

Cisco Switch

showfcns database detail vsan x

Example:

------------------------

VSAN:100   FCID:0x290000

------------------------

port-wwn(vendor)           :50:06:04:8c:4a:37:ab:cd(EMC)     

node-wwn:50:06:04:8c:4a:37:ab:cd

class                       :3

node-ip-addr                :0.0.0.0

ipa                         :ff ff ff ff ffff ff ff

fc4-types:fc4_features      :scsi-fcp 253

symbolic-port-name          :EMC SYMMETRIX 000287892468 SAF- 4cB EMUL B61F0000 1294AB11 3D1900 2C97C4 04.20.10 12:24 5671_081

symbolic-node-name          :EMC SYMMETRIX 000287892468 SAF- 4cB EMUL B61F0000 1294AB11 3D1900 2C97C4 04.20.10 12:24 5671_081

port-type                   :N

port-ip-addr                :0.0.0.0

fabric-port-wwn:20:07:00:05:73:f0:98:76

hard-addr                   :0x000000

permanent-port-wwn(vendor) :50:06:04:8c:4a:37:ab:cd (EMC)     

connected interface         :fc1/9

switch name (IP address)    :switch01 (1.2.3.4)

From the Name Server’s database, we can get the PWWN/NWWN, even the brand, device type and driver version information.

During the process of PLOGI to Name Server, the initiators (usually hosts) will issue an inquiry. According to the Zoning configuration, the Name Server replies with the nodes which can communicate with the initiators. If the Zoning is enabled, but there is no device in zones, the devices could not communicate with each other. If Zoning has been enabled yet, then all the nodes are in a default large zone. Whether the devices can talk with each other depends on the default zoning policy. 

Using the following commands to check the zoning configuration:

Brocade Switch

cfgactvshow

Example:

Effective configuration:                

cfg: Brocade                                   

zone:  VER093046                            

10:00:00:05:33:90:ab:7e

10:00:00:05:33:90:ab:7f

50:06:01:67:3e:a0:04:4a

50:06:01:6e:3e:a0:04:4a

Cisco Switch

showzoneset active vsan x

Example:

`show zoneset active vsan 100`

zoneset name Fabric_A vsan 100                                                           

zone name Mercury_A_VNX1408_SPA0_SPB1 vsan 100                             

* fcid 0x370000 [pwwn50:06:01:60:47:20:12:34]                              

* fcid 0x370100 [pwwn50:06:01:69:47:20:12:34]                              

* fcid 0x370200 [pwwn10:00:8c:7c:ff:21:ab:cd]

When creating zones, EMC recommends use PWWN as the members, and each zone contains only one initiator (an HBA port). This is to avoid the initiators interfere with each other.

Step 4&5.End-to-end PLOGI and PRIL

After the device finishes the FLOGI/PLOGI/SCR process, logins to the switch, and meanwhile the zoning configuration is correct, the initiator and target can communicate with each time at FC2 layer.

Initiator needs to initiate the PLOGI process to Target again. Once it is done, initiate the PRIL (Process Login) process to establish the infrastructure for upper layer protocol (ULP). Usually FC’s ULP is SCSI. In this phrase, the switch only transmits FC frames. It will not be involved in PLOGI/PRIL processes. 

From all of the above, we can see that a switch will participate in steps 1 to 3. If the host didn’t login to the storage, we should check the following items at the switch side:

1.    If the end device has been assigned the FC address (FLOGI is successful)

2.    If the end device has been registered to a Name Server (PLOGI to Name Server is successful)

3.    If the zoning configuration is correct and activated

4.    If there is packet loss on switch port

If all of the above items are Ok, then probably there is no problem with the switch. You may need to investigate on the hosts, storages and so forth (drives, firmware, and compatibility).