Review:
In the previous article, “How to Mount NFS (2),” you were introduced to the second method of NFS mounting, which is direct installation. The biggest difference between a direct installation and a predefined installation is that the former does not actively update its information in filesystems files, which is why we also call direct installation temporary. I will not repeat the specific operation here, if you are interested in can go to my last article. The links are as follows: How to Mount NFS(2)
To know the new:
Today I want to talk to you about NFS problem diagnosis, mainly about some inspection methods. Said a situation I met before, saying I do one day AIX system inspection, suddenly found a df – g command results showed that there was no start you didn’t, and the later found to be mounted NFS file system out of the question, for remote NFS file system after the repair, found again to use df – g, there is no problem. This is just to give you an idea, the next thing I want to talk about is the focus of this time.
If an NFS problem is found, you need to troubleshoot the problem from the following three directions: server, client, and network. The specific investigation ideas can be roughly divided into the following four aspects:
(1) Check that the network connection is normal. You can run the ping command to confirm the fault.
(2) Check whether the Inetd, PortMap and Biod processes are running properly on the client.
(3) Check that the file system has the correct mount point directory on the client.
(4) Run the following command to verify whether the server is running properly.
Check on the client:
Check whether the local NFS process can be used
lssrc -g nfs
Copy the code
Check whether the NFS server is available
/usr/bin/rpcinfo -p HOST name of the NFS server After the command is executed, if the server runs normally, a list of programs, such as the version number, protocol number, and port number, is displayedCopy the code
Check whether the daemon process of the NFS server is available
/usr/bin/rpcinfo -u NFS server mount /usr/bin/rpcinfo -u NFS server portmap /usr/bin/rpcinfo -u NFS server NFS If similar results are not displayed in the following figure, Note The server may be faultyCopy the code
Check the network connection of the NFS client
nfsstat -c
Copy the code
The value of the Retrans column indicates the number of retransmissions. If the value of the Retrans column divided by the value of the Call column on the client is greater than 5%, that is, retrans÷ Call ×100>5%, it indicates that retransmission is serious.
Check on the server:
Check whether the process is started
lssrc -g nfs
Copy the code
Check the exports file for your server
showmount -e
Copy the code
Check the network connection of the server
nfsstat -s
Copy the code
expand
Common NFS error information is as follows:
1. The hard installation and soft installation files are faulty.
The problem with accessing hard-mount and soft-mount remote files is different when there is a network or server problem. NFS server HOSTNAME Not Responding, still Trying If the server is not responding to a hardinstall request, NFS on the client displays the following information: NFS Server HOSTNAME Not responding, still Trying NFS on the client displays the following message: Connection timed out
2. Bad Sendreply message.
An insufficient network transfer cache may cause the following error message: nfs_server: bad Sendreply Use the smitty commodev command to increase the transfer cache by first selecting the adapter type and then increasing the transfer cache size.
3. The server does not respond.
You can diagnose this error (the server is not responding) using the methods described in Section 18.6.1. This error usually occurs when the NFS process is not started or has stopped running. If the mountd or NFSD process on the server is not started or has been stopped, an 1831-010 error message will be displayed when the client attempts to install an exported file system. If the RPC, mountd process is started and then dies, the client will receive this error message and resolve the problem as follows
(1) Telnet to the NFS server and log in to the system as the root user.
(2) Run the CD command to go to the /etc directory;
(3) Run stopsrc -g NFS command.
(4) Run stopsrc-s portmap command. (5) Run the rm-rf state sm sm.bak xtab rmtab command.
(6) Run startsrc -s portmap command;
(7) Run startsrc-g NFS command.
(8) Run exportfs-a; (9) Run the showmount e servername command, where servername is the host name of the NFS server.
4. The remote installation is incorrect.
(1)mount: .. Already mounted. The file system you are trying to install is already installed.
(2)mount .. Not found in /etc/filesystems. The specified file system does not match the directory name. If the mount command is executed with only the filesystem name or directory name parameter, the command will look for the record that the filesystem field or directory field matches the parameter in the /etc/filesystems file, and then install the NFS directory according to the description of the record. (3) mount.. Not in hosts database. On a Network that does not use the Network Information Service (NIS), this error message indicates that the host name used by the mount command is not in the /etc/hosts file. On an NIS network, this error message indicates that NIS cannot find the host name in the /etc/hosts file, or that the NIS ypbind process has terminated. If /etc/resolv.conf exists, use a nameserver for host name resolution because there may be problems in the specified database. Check the spelling and grammar of the mount command. If the command is spelled correctly, NIS is not running on the network, and the error occurs only because of the host name, you should check the records in the /etc/hosts file. If NIS is running on the network, run the following command to check whether the ypbind process is running properly: # ps -eflgrep ypbind If there is a record of the ypbind process, then run the rlogin command to login to the remote host. Or use the RCP command to copy some files to the remote host. If these commands fail, the ypbind process may stop running or be suspended. If the error occurs only because of the host name, you should check the records in the /etc/hosts file on the NIS server. (4)mount .. Server Not responding: Port mapper failure-RPC timed out. The NFS server you are trying to install may be powered off or not working properly, or the portMap process may be stopped or suspended, so the Inetd, portmap, and ypbind processes should be restarted. If you cannot login to the remote server using the rlogin command and the server is working properly, you should check the network connection by logging in to another remote host, and you should also check the network connection of the server. (5)1831-019 mount .. Server Not Responding: Program not registered. This means that the mount command makes contact with the portMap process, but the rpc.mountd process is not registered.
(6) mount: access denied. The host name of the client where the user resides is not listed in the export option for this file system on the NFS server. To get a list of exported file systems from the server, run the following command: # showmount -e host_name If the file system you want to install, or if the host name or netgroup name is not in the list of users for the file system, Log in to the server and check the appropriate file system records in the /etc/exports file. If a file system exists in the /etc/exports file but is not in the output of the showmount command, the mountd process may be faulty. The mountd process cannot parse records in the /etc/exports file, and it cannot find the directory, or the directory name is not a local directory that can be installed. If the /etc/exports file is correct and NIS is running on the network, you should check the ypbind process on the server, it may be down or suspended. (7) mount…. Permission denied. This information usually indicates that there may be a problem with authentication on the server. In the previous example, the user’s host name or user name was not in the export list on the server, the server could not recognize the ypbind process on the host, or the server could not accept the identity provided by the user. Check the /etc/exports file and the ypbind process on the server. If they are valid, in this case you can only change the hostname using the hostname command and try the mount command again. (8) the mount… Not a directory. The remote or local path is not a directory. After checking the spelling of the command, try executing the command on both the remote and local paths. (9) mount:… You are not allowed. A user must be root or a member of a system group to run the mount command because it affects all users on the machine using the file system. Only the root user and system group users can install or uninstall the NFS.
\