Hello All,

I had a problem on nfs, so I would like to ask all of you hopefully can help solve my problem. thanks,

I see the logs in / var / log / message on one client NFS (Canopus)


nfs: server 10.1.3.86 OK
nfs: server 10.1.3.86 not responding, still trying

the log is full with that error. I try ping from the canopus to centaury but no problem.

this is my server detail :

1. Server Centaury : running HP Cluster File System For linux versi 3.5.1
folder yang di share /docmgmt
SUSE LINUX Enterprise Server 9 (x86_64)
VERSION = 9
PATCHLEVEL = 3
kernel version = 2.6.5-7.252-smp
ip = 10.1.3.86


2. client NFS canopus : mounting on /data --- df -h = 10.1.3.86:/docmgmt /data
Red Hat Enterprise Linux Server release 5.3 (Tikanga)
kernel version = 2.6.18-128.el5PAE
ip = 10.1.15.101


and my canopus fstab configuration is here :

10.1.3.86:/docmgt /data nfs rw,rsize=32768,wsize=32768,hard,intr,noatime 0 0


Thanks in advance


Mansur

What is in /etc/exports on 10.1.3.86 (century)?

Do you have firewall (e.g. iptables) on century? If so are you allowing NFS connections (port 2049 usually)?

From canopus can you ping century? Can you "telnet 10.0.1.3.86 2049" and get a connection?

If you do "host 10.1.3.86" on canopus does it show the name is century (or is it in your /etc/hosts on canopus)? If you do "host century" on canopus does it show the IP is 10.1.3.86?

If you do the host commands on century for name canopus and its IP do you get the expected response?

Hi MensaWater,

Thanks for your response.

i'm sorry, there is a wrong of ip address of centaury. it must 10.1.3.32, 10.1.3.86 is ip virtual host used by cluster. ( i've 2 server cluster ).

here is /etc/exports in centaury :

/docmgt_system 10.1.21.0/24(rw,no_root_squash,sync)
/dms 10.1.3.114(rw,no_root_squash,sync)
/dms 10.1.3.16(rw,no_root_squash,sync)
/dms 10.1.3.120(rw,no_root_squash,sync)
/dms 10.1.21.0/24(rw,no_root_squash,sync)
/tcdata 10.1.3.22(rw,no_root_squash,sync)
/data 10.1.3.16(rw,no_root_squash,sync)
/docmgt 10.1.3.29(rw,no_root_squash,sync)
/docmgt 10.1.3.129(rw,no_root_squash,sync)
/docmgt 10.1.3.26(rw,no_root_squash,sync)
/docmgt 10.1.22.102(rw,no_root_squash,sync)
/docmgt 10.1.15.101(rw,no_root_squash,sync)
/docmgt 10.1.43.101(rw,no_root_squash,sync)
/docmgt 10.1.21.0/24(rw,no_root_squash,sync)
/docmgt 10.1.22.101(rw,no_root_squash,sync)
/docmgt 10.1.15.105(rw,no_root_squash,sync)
/email 10.1.3.110(rw,no_root_squash,sync)
/syslog 10.1.3.28(rw,no_root_squash,sync)


i don't have firewall on centaury and i have been allowing connection on port 2049.

from canopus, i can ping the centaury and if i try to telnet 10.0.1.3.32 2049 is no problem.

If i do "host 10.1.3.32" on canopus does it show the name is centuury and if i do "host century" on canopus does it show the IP is 10.1.3.32 also.



MensaWater,

in the cluster, i have 2 server. There are centaury and Atura. centaury is primary and atura is backup.

the real ip address of centaury is 10.1.3.32 dan atura is 10.1.3.33.

the cluster running virtual host with ip address 10.1.3.86 ( nfsexport2 ) and 10.1.3.87 ( nfsexport ), the member of that cluster is centaury and atura.

so, if the canopus ( with other server ) want to acces the /docmgmt, the request will provide by virtual host and the cluster will assign to Centaury ( primary ).

Are you exporting a filesystem from the cluster? That is to say will /docmgt filesystem move from centaury to aturi in a failover or does it simply become unavailable if centaury is donw? If so you probably need to specify the IP of the cluster rather than the one for centaury in canopus' fstab file. (You also have to setup of /etc/exports on aturi and may need to do other work in your cluster to make it run exportfs in a failover.)

It is unusual to use the same share multiple times on separate lines. I couldn't find anything that specifically prohibits it the way you're doing and posts I found on searching suggests it is allowed in one place and not allowed in another. You might want to try combining the lines various entries by putting line continuation so it looks like one big line even though it is on separate lines. ("man export" shows this.)

Does the mount on any host after 10.1.3.29 in the file work? If so you can ignore the above.

You might try doing ssh from canopus to centaury and run "who am i" to see what IP or host name it thinks your coming from. I've seen issues where there are multiple NICs and the server isn't using the one expected. You should also try it to the virtual IP from canopus.

Hi MensaWater,


I'am so sorry, thanks very much for your help


yes, i'm exsporting a filesystem from the cluster, that is /docmgt. FS /docmgt can move to atura in a failover.

i try ssh from canopus to centaury and to virtual ip of cluster , it's success.

MensaWater,


is there any effect if the interface is busy because of the transfer and receive on an interface centaury ?

becaus i look on Centaury interface Rx= 37GB and Tx 92GB