Finding devices based on a known ib interface (ibx/ehcax)

Use this procedure if the InfiniBand (ib) interface number is known, and the physical HCA port and attached physical switch port must be determined.

This procedure applies to IBM® GX HCAs.

Note: This procedure has some steps that are specific to operating system type (AIX® or Linux®). These steps have to do with querying the HCA device from the operating system. For AIX, the adapter is called ibax, where x is 0 - 3. For Linux, the adapter is call ehcax, where x is 0 - 3.

For example, if there is a problem with ib0, use the following procedure to determine the physical HCA port and physical switch port associated with the problem.

Record the ib interface number and server, for example, ib1 on c924f1ec09.
Log on to the server with the ib interface that is of interest.
From netstat, obtain the logical HCA GUID associated with the ib interface:
- For the AIX operating system, use netstat –I [ib interface]. You need to add leading zeros to bytes that are returned with single digits. You need the last 8 bytes of the address.
  Example results:
```
> netstat –I ib1

Name  Mtu   Network     Address              Ipkts Ierrs    Opkts Oerrs
Coll
ib1   65532  link#3      0.0.0.b.fe.80.0.0.0.0.0.1.0.2.55.0.10.24.d9.1
65     0        7     0     0
ib1   65532  192.168.9   192.168.9.65            65     0        7     0
0
```
  Therefore, a GUID of 0.2.55.0.10.24.d9.1 equals 00.02.55.00.10.24.d9.01.
- For the Linux operating system, use ifconfig [ib interface];
  Example results:
```
> ifconfig ib0 | grep inet6

inet6 addr: fe80::202:5500:1024:d900/64 Scope:Link
```
  Therefore, a GUID of 02:5500:1024:d900 equals 0002:5500:1024:d900 after you have added the leading zeroes.

Obtain the adapter device by performing the following actions:

For the AIX operating system, use the following command:

ibstat -p | grep -p "[1st seven bytes of GUID]" | grep iba

Example results:

> ibstat -p | grep -p "00.02.55.00.10.24.d9" | grep iba

IB PORT 1 INFORMATION (iba0)
IB PORT 2 INFORMATION (iba0)

The device is iba0.

For the Linux operating system, use the following command:

ibv_devinfo| grep –B1 "[1st seven bytes of GUID]" | grep ehca

Example results:

ibv_devinfo | grep –B1 "02:5500:1024:d9" | grep ehca

hca_id: ehca0

The device is ehca0.

Find the logical switch that is associated with the logical HCA for the interface.
Log on to the fabric management server.
Translate the operating system representation of the logical HCA GUID to the subnet manager representation of the GUID.
1. For GUIDs reported by AIX, delete the dots: 00.02.55.00.10.24.d9.00 becomes 000255001024d900
2. For GUIDs reported by Linux, delete the colons: 0002:5500:1024:d900 becomes 000255001024d900
Find the logical HCA GUID connection to the logical switch:
1. If the baseline health check has been run, use the following command. If it has not been run, go to step b.
```
grep –A 1 “0g *[GUID] *[port]” /var/opt/iba/analysis/baseline/fabric*links
```
2. If the baseline health check has not been run, query the live fabric by using the following command.
```
iba_report –o links | grep –A 1 “0g *[GUID] *[port]”
```
  Example results:
```
> grep –A 1 “0g 0x00025500103a7200* *1” /var/opt/iba/analysis/baseline/fabric*link

60g 0x000255001024d900   1 CA IBM G2 Logical HCA
<->  0x000255001024d902   2 SW IBM G2 Logical Switch 1
```
  The logical switch port is in the last line of the results of the query. This command tells you which logical switch attaches to the physical switch port.

Record the logical switch GUID. To obtain the name for the logical switch, enter the following command:

<->  [logical switch GUID]   [port] SW [logical switch name/IBnodeDescription]

Example results:

Logical switch 1; logical switch GUID = 0x0025501024d902

To find the physical location of the logical switch port, use the logical switch number and iba device found previously with the procedure: IBM GX HCA physical port mapping based on device number.
Example results:
```
iba0/ehca0 and logical switch 1 map to C65-T1
```

Find the physical switch connection to the logical switch:

If the baseline health check has been run, use the following command. If it has not been run, go to step b.
```
grep –A 1 “0g *[GUID]” /var/opt/iba/analysis/baseline/fabric*links
```

If the baseline health check has not been run, query the live fabric by using the following command.

iba_report –o links | grep –A 1 “0g *[GUID]”

Example results:

> grep –A 1 “0g * 0x00025500103a7202” /var/opt/iba/analysis/baseline/fabric*links

20g 0x000255001024d902   1 SW IBM G2 Logical Switch 1
<->  0x00066a00d90003d3   3 SW SilverStorm 9024 DDR GUID=0x00066a00d90003d3

Obtain the name and port for the switch. The physical switch port is in the last line of the results of the query. The name might have been given such that it indicates where the switch is physically located.
```
 <->  [switch GUID]   [port] SW [switch name/IBnodeDescription]
```
Example results:
```
Port 3 on switch SilverStorm 9024 DDR with a GUID of 0x00066a00d90003d3
```
This switch has not been renamed and is using the default naming convention, which includes the switch model and GUID.

Therefore, for ib0 in the server, the C65-T1 HCA port is attached to port 3 of SilverStorm 9024 DDR with a GUID of 0x00066a00d90003d3.