[LU-6340] DLC: /etc/rc.d/init.d/lnet doesn't configure lnet exactly as defined in lnet.conf Created: 06/Mar/15  Updated: 16/Jan/16  Resolved: 06/Jul/15

Status: Resolved
Project: Lustre
Component/s: None
Affects Version/s: Lustre 2.7.0
Fix Version/s: Lustre 2.8.0

Type: Bug Priority: Critical
Reporter: Sarah Liu Assignee: Amir Shehata (Inactive)
Resolution: Fixed Votes: 0
Labels: None
Environment:

lustre-b2_7 build #29

Issue Links:
Duplicate
is duplicated by LU-7474 Using lnet init script and lnetctl br... Resolved
Related
Severity: 3
Rank (Obsolete): 17752

 Description   

After running /etc/rc.d/init.d/lnet start, it doesn't configure lnet exactly as defined in /etc/sysconf/lnet.conf. Router is missing, net are not the exact same.

This is a minor issue, Amir has suggested 1 line change can fix this problem .

the lnet.conf looks like

[root@eagle-54vm3 modprobe.d]# more /etc/sysconfig/lnet.conf 
net:
    - net: lo
      nid: 0@lo
      status: up
      tunables:
          peer_timeout: 0
          peer_credits: 0
          peer_buffer_credits: 0
          credits: 0
    - net: tcp1
      nid: 192.168.201.180@tcp1
      status: up
      interfaces:
          0: eth1
      tunables:
          peer_timeout: 181
          peer_credits: 8
          peer_buffer_credits: 0
          credits: 256
route:
    - net: tcp
      gateway: 192.168.201.101@tcp1
      hop: 1
      priority: 0
      state: up
[root@eagle-54vm3 modprobe.d]#

after running lnet start check the system

[root@eagle-54vm3 modprobe.d]# lnetctl net show -v
net:
    - net: lo
      nid: 0@lo
      status: up
      tunables:
          peer_timeout: 0
          peer_credits: 0
          peer_buffer_credits: 0
          credits: 0
    - net: tcp
      nid: 10.100.4.198@tcp
      status: up
      tunables:
          peer_timeout: 180
          peer_credits: 8
          peer_buffer_credits: 0
          credits: 256
    - net: tcp1
      nid: 192.168.201.180@tcp1
      status: up
      interfaces:
          0: eth1
      tunables:
          peer_timeout: 181
          peer_credits: 8
          peer_buffer_credits: 0
          credits: 256
[root@eagle-54vm3 modprobe.d]# lnetctl route show -v
[root@eagle-54vm3 modprobe.d]#


 Comments   
Comment by Andreas Dilger [ 06/Mar/15 ]

Amir, can you please provide a patch, in case we have another 2.7 RC5.

Comment by Amir Shehata (Inactive) [ 06/Mar/15 ]

The issue here is that "lctl network up" is used to bring up LNet, but that configures everything in module parameters. The fix is to use lnetctl to configure LNet only if the proper lnet YAML configuration file is present. This way the script remains backward compatible.

Comment by Gerrit Updater [ 06/Mar/15 ]

Amir Shehata (amir.shehata@intel.com) uploaded a new patch: http://review.whamcloud.com/14000
Subject: LU-6340 lnet: LNet startup script fix
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: e3b4da8c93b7421802e3b8d03c6ef2b913eb56c5

Comment by Gerrit Updater [ 18/Mar/15 ]

Oleg Drokin (oleg.drokin@intel.com) merged in patch http://review.whamcloud.com/14000/
Subject: LU-6340 lnet: LNet startup script fix
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: f7259a01bb3f18f571d6ed8ecf7c95e1f9521bbb

Generated at Sat Feb 10 01:59:22 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.