This is due to a bug that was fixed introduced in build 125 and fixed in build 126.

   - Garrett

Jens Elkner wrote:
Hi,

I'm trying to setup [NFS] loadbalancing on X4440s via nge0 and nge1
using Nevada b125.

The problem here is, that after an reboot the svc:/network/physical:default
goes into maintainance mode, probably because it does not add nge1 to
the aggregation:

# dladm show-aggr -x
LINK        PORT           SPEED DUPLEX   STATE     ADDRESS            PORTSTATE
aggr1       --             1000Mb full    up        0:21:28:44:2d:b2   --
            nge0           1000Mb full    up        0:21:28:44:2d:b2   attached

# dladm show-aggr -L
LINK        PORT         AGGREGATABLE SYNC COLL DIST DEFAULTED EXPIRED
aggr1       nge0         yes          yes  yes  yes  no        no

# dladm show-aggr -P
LINK            POLICY   ADDRPOLICY           LACPACTIVITY  LACPTIMER   FLAGS
aggr1           L4       auto                 active        short       -----

#  cat /etc/svc/volatile/network-physical:default.log
[ Oct 29 23:48:47 Enabled. ]
[ Oct 29 23:48:48 Executing start method ("/lib/svc/method/net-physical"). ]
[ Oct 29 23:48:48 Timeout override by svc.startd.  Using infinite timeout. ]
ifconfig: SIOCSLIFNAME for ip: aggr1: no such interface
add net default: gateway 192.168.1.1: Network is unreachable
[ Oct 29 23:48:49 Method "start" exited with status 96. ]

# ls -al /dev
total 304
drwxr-xr-x 257 root     sys          257 Oct 29 22:44 .
drwxr-xr-x  29 root     root          32 Oct 29 22:35 ..
lrwxrwxrwx   1 root     root          30 Oct 28 23:20 aggr -> 
../devices/pseudo/cl...@0:aggr
lrwxrwxrwx   1 root     root          30 Oct 28 23:20 aggr1 -> 
../devices/pseudo/a...@0:aggr1
..
lrwxrwxrwx   1 root     root          29 Oct 28 23:09 nge -> 
../devices/pseudo/cl...@0:nge
lrwxrwxrwx   1 root     root          38 Oct 28 23:09 nge0 -> 
../devices/p...@0,0/pci10de,c...@8:nge0
lrwxrwxrwx   1 root     root          38 Oct 28 23:09 nge1 -> 
../devices/p...@0,0/pci10de,c...@9:nge1
..

However, when I force device reconfiguration (touch /reconfigure),
the machines comes up with a working aggregation:

elkner.trex ~ > dladm show-aggr -x
LINK        PORT           SPEED DUPLEX   STATE     ADDRESS            PORTSTATE
aggr1       --             1000Mb full    up        0:21:28:44:2d:b2   --
            nge0           1000Mb full    up        0:21:28:44:2d:b2   attached
            nge1           1000Mb full    up        0:21:28:44:2d:b3   attached
elkner.trex ~ > dladm show-aggr -L
LINK        PORT         AGGREGATABLE SYNC COLL DIST DEFAULTED EXPIRED
aggr1       nge0         yes          yes  yes  yes  no        no
--          nge1         yes          yes  yes  yes  no        no

Problem is the same on both X4440 I'm currently testing and always
reproducable as described above.


2nd problem which comes and goes with certain S10/S11 releases:
nscd seems to be a little bit brain damaged - after a reboot one always
needs to login and pkill nscd before NIS+ users are able to login - strange!
(I guess this is also the problem, why some vfstabbed NFS are not mounted
and one needs to mount -a  manually).


Any hints, what's going wrong here?

Regards,
jel.

_______________________________________________
networking-discuss mailing list
[email protected]

Reply via email to