What is the meaning of the error, "Bind failed: port already in use."?
This means another process is using the DHCP server port. Most likely, you have a BOOTP server already running on your system. Check to make sure that bootps is commented out in/etc/inetd.conf. Also, make sure that you don't have a BOOTP process running (for example, ps -ef | grep bootp).
How do I specify a WINS server for NT and Windows `95 clients?
WINS server is the same as the NetBIOS Name Server (NBNS). You can configure NBNS using thexjoin GUI.
How can I find out more information about my lease if I am using Windows `95 or Windows NT?
For Windows `95 use thewinipcfg utility. The winipcfg utility allows you to examine lease information such as the leased IP address, netmask, default gateway, expiration time, renewal time, and so on. You can also use it to manually renew or release an address.For Windows NT, use the
ipconfig utility. You can get DHCP information by executing the command ipconfig /all in a DOS-box.
How can I force the client to accept parameters offered by the server?
You can't. The server has no authority over the client's configuration. It is up to the client to accept the offered parameters. The client may choose to accept some of the parameters or reject the offer completely.
How do I create an exact replica of my DHCP server on another machine?
To create a mirror server, you basically need to transfer all relevant configuration information and the "ownership" of leases from one server to another. Suppose you want to transfer DHCP from Server A to Server Z, you can do so by following these steps:
- Transfer the
netmasks,server.pcy,dhcptagsanddhcpcapfiles to the new server. - Transfer the
netsandnamepoolfiles to the new server. Modify them by replacing references to the old server with references to the new server. If you used Server A's IP address, change it to Server Z's IP address. If you used Server A's hostname, change it to Server Z's hostname. - On the new server, remove the
db*.hshanddb*.btrdatabase files, if they exist. - Transfer the
dbf.btranddbj.hshdatabase files to the new server. - On Server A (old server), dump all the lease information from the dynamic database into a temporary file:
- Modify the
tempfile and replace all references to the old server's IP address with references to the new server's IP address. - Transfer the
tempfile to the new server, and re-create the database:
# jdbmod temp_file
After performing the above steps, you can only run one of the two servers. The two servers MUST NOT run concurrently.
How do I get the server to re-initialize without stopping the server?
If you use the xjoin GUI to configure the server, the changes take effect immediately. However, if you manually modify the configuration files (i.e., dhcpcap, nets, etc.), you may need to notify the server of the changes. You can force the server to re-initialize by sending it a HUP signal:
# kill -HUP <pid-of-joind>
What sort of "signals" can I send to the server?
The joind process accepts three UNIX signals that you can send via the kill command.
| Signal | Description |
|---|---|
| HUP | Re-initialize the server |
| USR1 | Flush all cached information to the disk. If the parameter, "Auto Sychronize Database" is set to True, there is no need to send this signal to the server. The server will automatically write lease information to the disk after each transaction. |
| USR2 | Re-initialize the FLEXlm license database. This will force the server to acquire the license for all active leases only. License for all expired leases will be relinquished. Typically, you do not need to send this signal to the server. By default, the server will re-initialize the FLEXlm license database every hour. However, you can use this signal to force the server to re-initialize FLEXlm licenses at any time. |
How do I make the client's IP address its default gateway address?
In some environments it may be desirable to have the client's default gateway the same as its IP address. This may occur in a flat network or in networks where secondary addressing support is necessary. To make the server return the client's IP address as its default gateway, simply specify the IP address 0.0.0.0 as the default gateway.
How can I restart with a clean database? Can I delete all the database files (db*) and start from scratch again?
Deleting all the database files can certainly erase all records of lease information. However, that approach is not recommended. If you suspect that the JOIN database is somehow corrupted or you simply want to remove old, expired leases you can follow these steps to "dump" existing records and then re-build the database with active (non-expired) leases:
- Stop the
joindserver. - Dump lease information into a temporary file:
- Make a backup copy of all the database files (in case you need them back).
- Remove all database files (
db*). If you are using the "Restrict to Known MAC address" feature, you need to preserve database J (dbj.hsh). - Reconstruct the database with information you "dumped" earlier:
- Restart the
joindserver.
# jdbdump > temp_file
# jdbmod temp_file
You may ignore the warning messages about expired leases not added to the database. When it is done, a new set of database files will be created with only the active leases.
If you really want to empty out the database, you can do the following:
- Stop the
joindserver. - Delete all entries from the database:
# jdbdump | jdbmod -d
How do I get DHCP clients on one subnet talk to a DHCP server on another subnet? Or what is a relay agent?
When DHCP clients and DHCP server are on different networks, separated by a router, you need a BOOTP/DHCP "relay agent". A relay agent routes BOOTP/DHCP packets between clients and server that reside on different networks. In most cases, your router should have a relay agent already built in. To relay BOOTP/DHCP packets, you need to configure the router with information about the location of the DHCP server. For some routers, this is known as the "helper address" which should simply be the IP address of the DHCP server.
What is the meaning of the message, "BOOTP from HW address %s: ignored because reply recently sent"?
You get this error message if a BOOTP client sends multiple requests in a short time frame. The server has a time window where it'll respond to a BOOTP client only once during that time period. This time period is specified by the parameter, "Provisional Time To Live". By default, it is 60 seconds. This means that once a BOOTP reply is sent, the server will not send out another BOOTP reply to the same client within 60 seconds. To reduce the chance of getting this warning message, reduce the provisional time-to-live to 3-5 seconds.
What is the meaning of the message, "received short packet length %d bytes from %s: discarded"?
You get this message if the server receives a packet with a packet size smaller than the recommended minimum packet size. Some early implementation sent out BOOTP request with a less than minimum packet size, for example, Macintosh OpenTransport 1.0. By default, the server will ignore these illegal requests. If you want the server to accept these requests, lower the parameter, "minimum packet size" to a value acceptable to the client.
What is the meaning of the message, "DB is not writable: locked by another process, stop other process or remove the lock file %s"?
Upon startup, the server acquires exclusive access to the lease database files. It does so by creating a group of lock files (rwlock*). Whenever the server comes up, it checks to see if these files exist. If they exist that means someone else is trying to access these files. The server will then display the above message to let the user know that exclusive access to these files is possible.This condition can occur if the server is terminated without a graceful shutdown. These are some of the possible causes for an ungraceful shutdown:
- power failure
- a kill -9 signal is used to kill the server
- bug in the server code resulting in a core dump
To restart the server, simply remove all
rwlock* files from the database directory (default: /var/join) and run joind again. What is the meaning of the message, "mask %s on interface %s disagrees with netmasks database %s"?
This warning message typically signals that there is an error in the netmasks file. The first netmask value describes the netmask that is configured for the interface using the ifconfig command. The second netmask value is what the server thinks should be the correct netmask for that interface based on information from the netmasks file. To correct this problem, make sure that your netmasks file describe the proper netmasks for the interface.