[lustre-discuss] lctl replace_nids example

[Patricia Santos Marco]

Hello,


Recently a new gigabit network has been added our lustre servers and we
want  to add this network to the lustre system.

 In Lustre 2.4 there is a "lctl replace_nids" command that in theory
allows  to change the NIDs without running --writeconf. However I don't
find any example to how I can use this command. Anybody have used it for
this purpose?

For more info:

We added the new interface in lnet configuration (tcp0):
/etc/modprobe.d/lustre_interface.conf
options lnet networks=o2ib0(ib0),tcp0(eth2)

and we see both interfaces:
 lnetctl net show
net:
    - net: lo
      nid: 0 at lo
      status: up
    - net: o2ib
      nid: 192.168.2.252 at o2ib
      status: up
      interfaces:
          0: ib0
    - net: tcp
      nid: 192.168.3.252 at tcp
      status: up
      interfaces:
          0: eth2

[root at cmds ~]# lctl list_nids
192.168.2.252 at o2ib
192.168.3.252 at tcp

However new clients aren't able to connect lustre servers via the new
interface.

 LustreError: 15c-8: MGC192.168.3.252 at tcp: The configuration from log
'LUSTRE-client' failed (-2). This may be the result of communication errors
between this node and the MGS, a bad configuration, or other errors. See
the syslog for more information.

Thanks!




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

Patricia Santos Marco

HPC research group System Administrator

Instituto de Biocomputación y Física de Sistemas Complejos (BIFI)

Universidad de Zaragoza

e-mail: psantos at bifi.es <artginer at bifi.es>

phone: (+34) 976762992

http://bifi.es/~patricia/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-discuss-lustre.org/attachments/20160721/7e772ba1/attachment.htm>