EV_AGENT[2915]:

[ehab_4]

Hi all,
we faced these problems in our hp clustered nodes
EV_AGENT[2915]: Agent cannot communicate with the storage system
cmgmsd[3145]: Unable to accept a connection (activation of keep alive option on socket failed): Invalid argument
moihp2 EV_AGENT[2915]: Agent Main --Estream construct failed. Err: RWSockErr: in RWSocket::accept: NOBUFS
inetd[1673]: vnetd/tcp: accept: No buffer space available
Aug 22 20:32:58 moihp2 EV_AGENT[2915]: Agent Main --Estream construct failed. Err: RWSockErr: in RWSocket::accept: NOBUFS
then the nodes hanged
we had rebooted both nodes then things came normal
are there any idea what would be the problem
thank you.

[melvyn burnard]

seems like yuou may have had memory issues:

moihp2 EV_AGENT[2915]: Agent Main --Estream construct failed. Err: RWSockErr: in RWSocket::accept: NOBUFS
inetd[1673]: vnetd/tcp: accept: No buffer space available

[skt_skt]

i would also suspect the mem usage. Did you collect the MEM/CPU/SWAP statistics when the server got hung(Perfiew). Any thing special in syslog.log

whas is reporting from below command.
kmtune|egrep "dbc|shmmax"

I would say collect output of "kmeminfo" when you have a similar situation whihc can be a little helpful about the current memory allocation

[ehab_4]

how about the cluster is there any problem of the cluster componants like the switch or the network cards