Page: 1, 2, 3  Next

please show Infiniband devices also..
ricoshay
Status: Curious
Joined: 18 Jul 2014
Posts: 9
Reply Quote
:: Code ::
[r**t@bigpingu01 ~]# inxi -nN
Network:   Card-1: Intel 82576 Gigabit Network Connection driver: igb
           IF: eth0 state: up speed: 1000 Mbps duplex: full mac: 00:25:90:4e:9a:40
           Card-2: Intel 82576 Gigabit Network Connection driver: igb
           IF: eth1 state: down mac: 00:25:90:4e:9a:41

[r**t@bigpingu01 ~]# dmesg | grep Mell
mlx4_core: Mellanox ConnectX core driver v1.1 (Dec, 2011)
mlx4_en: Mellanox ConnectX HCA Ethernet driver v2.0 (Dec 2011)
<mlx4_ib> mlx4_ib_add: mlx4_ib: Mellanox ConnectX InfiniBand driver v1.0 (April 4, 2008)

[r**t@bigpingu01 ~]# ip a l
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP qlen 1000
    link/ether 00:25:90:4e:9a:40 brd ff:ff:ff:ff:ff:ff
    inet6 fe80::225:90ff:fe4e:9a40/64 scope link
       valid_lft forever preferred_lft forever
3: eth1: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN qlen 1000
    link/ether 00:25:90:4e:9a:41 brd ff:ff:ff:ff:ff:ff
4: ib0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 2044 qdisc pfifo_fast state UP qlen 256
    link/infiniband 80:00:00:48:fe:80:00:00:00:00:00:00:00:25:90:ff:ff:07:c6:a9 brd 00:ff:ff:ff:ff:12:40:1b:ff:ff:00:00:00:00:00:00:ff:ff:ff:ff
    inet 172.22.44.201/24 brd 172.22.44.255 scope global ib0
    inet6 fe80::225:90ff:ff07:c6a9/64 scope link
       valid_lft forever preferred_lft forever
5: ib1: <BROADCAST,MULTICAST> mtu 4092 qdisc noop state DOWN qlen 256
    link/infiniband 80:00:00:49:fe:80:00:00:00:00:00:00:00:25:90:ff:ff:07:c6:aa brd 00:ff:ff:ff:ff:12:40:1b:ff:ff:00:00:00:00:00:00:ff:ff:ff:ff
6: bond0: <BROADCAST,MULTICAST,MASTER> mtu 1500 qdisc noop state DOWN
    link/ether 00:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff
7: eth0.228@eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP
    link/ether 00:25:90:4e:9a:40 brd ff:ff:ff:ff:ff:ff
    inet 172.22.28.46/24 brd 172.22.28.255 scope global eth0.228
    inet6 fe80::225:90ff:fe4e:9a40/64 scope link
       valid_lft forever preferred_lft forever
8: ib0.8100: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 4092 qdisc pfifo_fast state UP qlen 256
    link/infiniband 80:00:00:4c:fe:80:00:00:00:00:00:00:00:25:90:ff:ff:07:c6:a9 brd 00:ff:ff:ff:ff:12:40:1b:81:00:00:00:00:00:00:00:ff:ff:ff:ff
    inet 10.10.100.1/24 brd 10.10.100.255 scope global ib0.8100
    inet6 fe80::225:90ff:ff07:c6a9/64 scope link
       valid_lft forever preferred_lft forever
9: ib0.8101: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 2044 qdisc pfifo_fast state UP qlen 256
    link/infiniband 80:00:00:4d:fe:80:00:00:00:00:00:00:00:25:90:ff:ff:07:c6:a9 brd 00:ff:ff:ff:ff:12:40:1b:81:01:00:00:00:00:00:00:ff:ff:ff:ff
    inet 10.10.101.1/24 brd 10.10.101.255 scope global ib0.8101
    inet6 fe80::225:90ff:ff07:c6a9/64 scope link
       valid_lft forever preferred_lft forever
10: ib0.8110: <BROADCAST,MULTICAST> mtu 4092 qdisc pfifo_fast state DOWN qlen 256
    link/infiniband 80:00:00:4e:fe:80:00:00:00:00:00:00:00:25:90:ff:ff:07:c6:a9 brd 00:ff:ff:ff:ff:12:40:1b:81:10:00:00:00:00:00:00:ff:ff:ff:ff
    inet 10.10.110.1/24 brd 10.10.110.255 scope global ib0.8110

Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 4127
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
this is not the place to file inxi issue reports.

code.google.com/p/inxi/issues/list

I'm not familiar with infiniband so you will have to explain what it is and why it should be included.

Also, run, with a fully upgraded inxi (2.1.28) this command: inxi -xx@ 14

which will upload all the system info I need to actually implement a feature.
Back to top
ricoshay
Status: Curious
Joined: 18 Jul 2014
Posts: 9
Reply Quote
InfiniBand is a switched fabric computer network communications link used in high-performance computing and enterprise data centers. Why we would like to see the Infiniband interfaces with 'inxi' is to easily identify devices and create kickstart scripts for auto deployment of server nodes. It is not critical to be able to see the Infiniband devices if it's to much trouble. We just thought that it would be nice to have. We'll be using 'inxi' anyway.

Thanks for the reply
Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 4127
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
I didn't say it was too much trouble, I just wanted to know why the feature was wanted, I take the simple position that if an inxi user can show a valid reason for a feature being added, which you have done now, it will be added if it's feasible.

I'll still need the: inxi -xx@14
data in order to implement the feature, that gives me all the working files about sys info that I can plug into the logic to emulate the systems and features.

I see one debugger upload, If this is your data, let me know: inxi-bigping....

Update: looks like it's the right one, thanks, I'll take a look. Always happy to support real servers and the needs of real sys admins with inxi.

Thanks.
Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 4127
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
I can get tentative but not perfect support.

to test this, update inxi like this:

inxi -! 11

this will update it from branches/one inxi, which has an attempt to add infiniband support.

It's too complicated for me to emulate this locally unfortunately.

Then test it locally like this: inxi -Iixxx

-I shows what should be 2.1.28-1-b1 to make sure you updated the right version.

You should now with inxi -nxxx see most of the infiniband data.

And I think with -ixxx you should see the networking data. Note that I believe -i would have shown this no matter what already, unless I'm mistaken.

note that I do not attempt to link the output of the -i lines with the -n lines, it's too complicated, so those two things are not connected internally.

Because your sample shows two ids, ib0 and ib1 on the same busid, inxi has to dump the second and more from that to get the rest of the data, so as is now, raw, you will only get card id from the first detected id, ib0, otherwise it won't work.

I can't guarantee this will work because I can't test it, but it's a start.

Post: inxi -ixxxIz

here, to hide the nic ids etc, after confirming that the mac address etc appears as expected.

inxi uses lspci -v to get the busid, then it uses the busid to query /sys to get the other data about the card, like speed, duplex (these two don't seem present on infiniband, duplex I can see why not), mac address, and state.

See how that goes.
Back to top
ricoshay
Status: Curious
Joined: 18 Jul 2014
Posts: 9
Reply Quote
This is great! Thanks a lot! The IB adapter has two ports and is probably the reason why ib0 is detected first. Port ib0 is up and ib1 is down.

:: Code ::
[r**t@bigpingu01 etc]#  inxi --version
inxi 2.1.28-01-b1 (2014-07-21)

## Here's the output from: inxi -ixxxlz

[r**t@bigpingu01 etc]# inxi -ixxxlz
Network:   Card-1: Mellanox MT26428 [ConnectX VPI PCIe 2.0 5GT/s - IB QDR / 10GigE]
           driver: mlx4_core v: 1.1 bus-ID: 02:00.0 chip-ID: 15b3:673c
           IF: ib0
ib1 state: N/A speed: N/A duplex: N/A mac: N/A
           Card-2: Intel 82576 Gigabit Network Connection
           driver: igb v: 4.0.1-k ports: ec00 Root bus-ID: 05:00.0 chip-ID: 8086:10e7
           IF: eth0 state: up speed: 1000 Mbps duplex: full mac: <filter>
           Card-3: Intel 82576 Gigabit Network Connection
           driver: igb v: 4.0.1-k ports: e880 Root bus-ID: 05:00.1 chip-ID: 8086:10e7
           IF: eth1 state: down mac: <filter>
           WAN IP: <filter> IF: bond0 ip: N/A ip-v6: N/A
           IF: ib0 ip: <filter> ip-v6: <filter> IF: ib1 ip: N/A ip-v6: N/A IF: eth0 ip: N/A ip-v6: <filter>
           IF: eth1 ip: N/A ip-v6: N/A IF: ib0.8100 ip: <filter> ip-v6: <filter>
           IF: eth0.228@eth0 ip: <filter> ip-v6: <filter> IF: ib0.8110 ip: <filter> ip-v6: N/A
           IF: ib0.8101 ip: <filter> ip-v6: <filter>
Partition: ID-1: / size: 50G used: 16G (34%) fs: ext4 dev: /dev/dm-0 label: N/A
           ID-2: /boot size: 485M used: 54M (12%) fs: ext4 dev: /dev/sda1 label: N/A
           ID-3: /home size: 405G used: 33G (9%) fs: ext4 dev: /dev/dm-2 label: N/A
           ID-4: swap-1 size: 16.90GB used: 0.00GB (0%) fs: swap dev: /dev/dm-1 label: N/A

Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 4127
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
Looks like I was close, but didn't quite get the extra data for the ib0, and there's a line break that needs to be cleaned out.

I'll upload a new branches version today to see if I can get that handled.

There is a way to also get the two ports on the ib device handled, but it's tricky, but may be worth trying, I'll see if I have time.

Sorry for lack of clarity, I wanted -l (Eye) not el, lol, but it's impossible to read them, but your output is better since all I really want to make sure of is that you got the right branch version, I'll ask for how you gave the data instead in the future since it is not ambiguous, ie:

inxi --version
inxi <new features>
Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 4127
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
Update: inxi -! 11

then try it again, this is 2.1.28-02-b1

This version should correct the failed id, and should fill in the id type data like state, mac address, etc.

Also, and this will handle all devices that may report more than one id per busid, it will now, when advanced mode is used -n/-i, also show one full output sequence each id as well (ie, one for ib0, one for ib1, with state up for ib0, state down for ib1, and so on). That wasn't as hard as I thought it would be to do.

Please confirm it's working with output, and once I see it's working, I'll put this into new stable.

The good thing about this update is that it handles all possible scenarios with more than one network port per pcibusid device, not just infiniband. Not for usb, but that's fine.
Back to top
ricoshay
Status: Curious
Joined: 18 Jul 2014
Posts: 9
Reply Quote
To cut a long story short the mac adress (Port GUID) of a IB port has 20 octects which differs from an ethernet port which has six octects. When using IPoIB the IP6 adress is based on this Port GUID.

# ‘MAC addresses are truncated for Infiniband’ (Puppet Labs) shows a resolved issue.
https://projects.puppetlabs.com/issues/1415

# IB ports illustrated
:: Code ::

[r**t@bigpingu01 ~]# ibstat
CA 'mlx4_0'
   CA type: MT26428
   Number of ports: 2
   Firmware version: 2.9.1000
   Hardware version: b0
   Node GUID: 0x002590ffff07c6a8
   System image GUID: 0x002590ffff07c6ab
   Port 1:
      State: Active
      Physical state: LinkUp
      Rate: 40
      Base lid: 23
      LMC: 0
      SM lid: 16
      Capability mask: 0x02510868
      Port GUID: 0x002590ffff07c6a9
      Link layer: InfiniBand
   Port 2:
      State: Active
      Physical state: LinkUp
      Rate: 40
      Base lid: 25
      LMC: 0
      SM lid: 16
      Capability mask: 0x02510868
      Port GUID: 0x002590ffff07c6aa
      Link layer: InfiniBand

[r**t@bigpingu01 ~]# ibv_devinfo
hca_id:   mlx4_0
   transport:         InfiniBand (0)
   fw_ver:         2.9.1000
   node_guid:      0025:90ff:ff07:c6a8
   sys_image_guid:   0025:90ff:ff07:c6ab
   vendor_id:         0x02c9
   vendor_part_id:      26428
   hw_ver:         0xB0
   board_id:         SM_1081010B01
   phys_port_cnt:      2
      port:   1
         state:      PORT_ACTIVE (4)
         max_mtu:      4096 (5)
         active_mtu:   4096 (5)
         sm_lid:      16
         port_lid:      23
         port_lmc:      0x00
         link_layer:      InfiniBand

      port:   2
         state:      PORT_ACTIVE (4)
         max_mtu:      4096 (5)
         active_mtu:   4096 (5)
         sm_lid:      16
         port_lid:      25
         port_lmc:      0x00
         link_layer:      InfiniBand

[r**t@bigpingu01 ~]# ibstatus
Infiniband device 'mlx4_0' port 1 status:
   default gid:        fe80:0000:0000:0000:0025:90ff:ff07:c6a9
   base lid:            0x17
   sm lid:               0x10
   state:            4: ACTIVE
   phys state:    5: LinkUp
   rate:               40 Gb/sec (4X QDR)
   link_layer:            InfiniBand

Infiniband device 'mlx4_0' port 2 status:
   default gid:    fe80:0000:0000:0000:0025:90ff:ff07:c6aa
   base lid:            0x19
   sm lid:       0x10
   state:       4: ACTIVE
   phys state:    5: LinkUp
   rate:               40 Gb/sec (4X QDR)
   link_layer:            InfiniBand

:: Code ::

# [b]output from inxi[/b]

[r**t@bigpingu01 ~]# inxi --version
inxi 2.1.28-02-b1 (2014-07-22)
[r**t@bigpingu01 ~]# inxi -Nx
Network:   Card-1: Mellanox MT26428 [ConnectX VPI PCIe 2.0 5GT/s - IB QDR / 10GigE]
           driver: mlx4_core v: 1.1 bus-ID: 02:00.0
           Card-2: Intel 82576 Gigabit Network Connection driver: igb v: 4.0.1-k ports: ec00 Root bus-ID: 05:00.0
           Card-3: Intel 82576 Gigabit Network Connection driver: igb v: 4.0.1-k ports: e880 Root bus-ID: 05:00.1
[r**t@bigpingu01 ~]# inxi -ixxxIz
Network:   Card-1: Mellanox MT26428 [ConnectX VPI PCIe 2.0 5GT/s - IB QDR / 10GigE]
           driver: mlx4_core v: 1.1 bus-ID: 02:00.0 chip-ID: 15b3:673c
           IF: ib0
ib1 state: N/A speed: N/A duplex: N/A mac: N/A
           Card-2: Intel 82576 Gigabit Network Connection
           driver: igb v: 4.0.1-k ports: ec00 Root bus-ID: 05:00.0 chip-ID: 8086:10e7
           IF: eth0 state: up speed: 1000 Mbps duplex: full mac: <filter>
           Card-3: Intel 82576 Gigabit Network Connection
           driver: igb v: 4.0.1-k ports: e880 Root bus-ID: 05:00.1 chip-ID: 8086:10e7
           IF: eth1 state: down mac: <filter>
           WAN IP: <filter> IF: bond0 ip: N/A ip-v6: N/A
           IF: ib0 ip: <filter> ip-v6: <filter> IF: ib1 ip: N/A ip-v6: N/A IF: eth0 ip: N/A ip-v6: <filter>
           IF: eth1 ip: N/A ip-v6: N/A IF: ib0.8100 ip: <filter> ip-v6: <filter>
           IF: eth0.228@eth0 ip: <filter> ip-v6: <filter> IF: ib0.8110 ip: <filter> ip-v6: N/A
           IF: ib0.8101 ip: <filter> ip-v6: <filter>
Info:      Processes: 449 Uptime: 169 days Memory: 1568.5/32099.3MB
           Init: Upstart v: 0.6.5 runlevel: 3 default: 3 Gcc sys: 4.4.7
           Client: Shell (bash 4.1.21 running in tty 1) inxi: 2.1.28-2-b1
[r**t@bigpingu01 ~]#

Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 4127
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
I thought I had it working, oh well.

I'll review the fix again to see why it didn't work. Basically each infiniband port once this works will show as a separate card/device in inxi output, there's no other way to do it. And the state/mac address will be filled out correctly, I thought I had it working but I guess not.
Back to top
Display posts from previous:   
Page: 1, 2, 3  Next
All times are GMT - 8 Hours