This document covers the following topics:
The following table lists available stubs and gives an overview of available features and supported transport methods.
Stub | Language | Transport Methods | More Information |
---|---|---|---|
Jaci | Java | TCP /SSL | See EntireX Java ACI. |
broker.s[o|l] | C | TCP / SSL | See below. |
The Broker stub can use TCP/IP and SSL. The term "SSL" in this section refers to both SSL (Secure Sockets Layer) and TLS (Transport Layer Security).
To use TCP/IP
Optional: set the timeout, see Setting the Timeout for the Transport Method.
The Broker stub requires the IP address and the TCP port number (if the Broker's default TCP port number 1971 cannot be used) for each BROKER-ID. Either add an entry in the Domain Name System (DNS) or modify your local hosts and services tables. See Modifying the Hosts and Services Tables.
You can check whether the Broker has already been added to your DNS with the command:
ping <broker-id>
for example: ping ETB001. If a message such as "...is alive" or "Reply from ..." is displayed (the text displayed varies depending on your ping implementation), the name is known to your DNS and the host where the Broker is running is reachable. However, this does not necessarily mean that the Broker is active.
ACI applications can use Secure Sockets Layer/Transport Layer Security (SSL/TLS) as the transport medium. The term "SSL" in this section refers to both SSL and TLS. ACI-based clients or servers are always SSL clients. The SSL server can be either the EntireX Broker or the Broker SSL Agent. For an introduction see SSL/TLS, HTTP(S), and Certificates with EntireX in the platform-independent Administration documentation.
With the Broker ACI, the SSL parameters (e.g. certificates) are provided with the function SETSSLPARMS
.
To use SSL
To operate with SSL, certificates need to be provided and maintained. Depending on the platform, Software AG provides default certificates, but we strongly recommend that you create your own. See SSL/TLS Sample Certificates Delivered with EntireX in the EntireX Security documentation.
Specify the Broker ID, using one of the following styles:
URL Style, for example:
ssl://localhost:2010
Transport-method Style, for example:
ETB024:1609:SSL
If no port number is specified, port 1958 is used as default.
Specify SSL parameters in the second parameter, for example:
'broker' etbcb "VERIFY_SERVER=N&TRUST_STORE=c:\\certs\\CaCert.pem"
If the SSL client checks the validity of the SSL server only, this is known as one-way SSL. The mandatory trust_store
parameter
specifies the file name of a keystore that must contain the list of trusted certificate authorities for the certificate of
the SSL server.
By default a check is made that the certificate of the SSL server is issued for the hostname specified in the Broker ID.
The common name of the subject entry in the server's certificate is checked against the hostname. If they do not match, the
connection will be refused.
You can disable this check with SSL parameter verify_server=no
.
If the SSL server additionally checks the identity of the SSL client, this is known as two-way SSL. In this case the SSL server requests a client certificate
(the parameter verify_client=yes
is defined in the configuration of the SSL server).
Two additional SSL parameters must be specified on the SSL client side: key_store
and key_passwd
.
This keystore must contain the private key of the SSL client. The password that protects the private key is specified with
key_passwd
.
The ampersand (&) character cannot appear in the password.
SSL parameters are separated by ampersand (&). See also SSL/TLS Parameters for SSL Clients.
Make sure the SSL server to which the ACI side connects is prepared for SSL connections as well. The SSL server can be EntireX Broker or Broker SSL Agent. See:
See table Using SSL/TLS with EntireX Components if SSL is required for other EntireX components.
The Broker stub requires the IP address and the SSL port number for each BROKER-ID
.
Either add an entry to the Domain Name System (DNS) or modify your local hosts and services tables.
See Modifying the Hosts and Services Tables.
If no port number is specified, port 1958 is used as default.
You can check whether the Broker has already been added to your DNS with a ping <broker-id>
command, for example:
ping ETB001
If a message such as "...is alive
" or "Reply from ...
" is displayed (the text displayed varies depending on your ping implementation),
the name is known to your DNS and the host where the Broker is running is reachable. However, this does not necessarily mean
that the Broker is active.
Take care if trace is switched on:
Warning: If stub tracing level is > 1, unencrypted contents of the send/receive buffers are exposed in the trace. |
Example on running the delivered ACI example:
C:\SoftwareAG\EntireX\examples\ACI\conversational\C\convSvr -blocalhost:1958:SSL -cACLASS -sASERVER -vASERVICE -x"VERIFY_SERVER=N&TRUST_STORE=C:\SoftwareAG\EntireX\etc\ExxCACert.pem"
The timeout settings of the transport layers are independent of the
broker's timeout settings, which are set by the application in the
WAIT
field of the broker ACI control block.
If the transport layer is interrupted, communication between the
Broker and the stub (i.e. client or server application) is interrupted as well.
To prevent a client from waiting for a Broker reply indefinitely, set a timeout
value for the transport method. The actual timeout for the procedure is then
the Broker timeout (which is set by the application in the
WAIT
of the broker ACI control block) plus this
value.
To set a transport timeout value
Set the environment variable ETB_TIMEOUT
:
Transport Timeout Value | Description |
---|---|
0 | Infinite wait for the application. |
n | Transport method waits additional time in seconds. A negative
value is treated as ETB_TIMEOUT=0 (infinite wait).
|
No environment variable defined | Transport method waits additional 20 seconds. |
See also UNIX Commands to Set the Environment Variables.
With transport method TCP/IP, the broker stub establishes one or more TCP/IP
connections to the brokers specified with BROKER-ID
.
These connections can
be controlled by the transport-specific CONNECTION-NONACT
attribute on the
broker side, but also by the transport-specific environment variable
ETB_NONACT
on the stub side.
If ETB_NONACT
is not 0, it defines the non-activity
time (in seconds) of active TCP/IP connections to any broker. See ETB_NONACT
under Environment Variables in EntireX.
Whenever the broker stub is called, it checks for the elapsed non-activity time
and closes connections with a non-activity time greater than the value defined with ETB_NONACT
.
Stubs capable of running in SRB mode do not support ETB_NONACT
handling.
Transport Non-activity Value | Description |
---|---|
0 | Infinite lifetime until application is stopped. |
n (seconds) | Transport connections with non-activity time greater than n will be closed. |
Nothing set | Infinite lifetime until application is stopped. |
The Hosts and Services tables are plain text files in directory /etc.
To add an entry to the hosts table
Add a line similar to the following to the local hosts file:
100.100.1.1 ETB226 # ETB test host name
To add an entry to the services table
Add lines similar to the following to the local services file:
ETB226 18492/tcp # ETB test host name ETB411 21234/tcp # ETB production host name
The broker stubs provide an option for writing trace files.
To switch on tracing for the broker stub
Before starting the client application, set the environment variable ETB_STUBLOG
:
Trace Level | Description | |
---|---|---|
0 | NONE |
No tracing. Switch tracing off. |
1 | STANDARD |
Traces initialization, errors, and all ACI request/reply strings. |
2 | ADVANCED |
Used primarily by system engineers, traces everything from level 1 and provides additional information, for example the Broker ACI control block, as well as information from the transports. |
3 | SUPPORT |
This is full tracing through the stub, including detailed traces of control blocks, message information, etc. |
Example:
ETB_STUBLOG
=2
If the trace level is greater than 1, unencrypted contents of the send/receive buffers may be exposed in the trace.
The trace file is created in the current directory. The name is pid
.etb where pid
is the process ID. If you want to write the trace file to a different location, set environment variable ETB_STUBLOGPATH
to the desired path.
See also UNIX Commands to Set the Environment Variables.
Remember to switch off tracing to prevent trace files from filling up your disk.
To switch off tracing for the broker stub
Set the environment variable ETB_STUBLOG
to NONE
or delete it.
Logging works for both TCP and SSL. Tracing is controlled
by the environment variable ETB_STUBLOG
.
csh or tcsh users use:
setenv ETB_STUBLOG tracelevel
bsh, ksh or bash users use:
ETB_STUBLOG=tracelevel; export ETB_STUBLOG
Possible values for tracelevel:
Trace Level | Description | |
---|---|---|
0 | NONE |
No tracing. Switch tracing off. |
1 | STANDARD |
Traces initialization, errors, and all ACI request/reply strings. |
2 | ADVANCED |
Used primarily by system engineers, traces everything from level 1 and provides additional information, for example the Broker ACI control block, as well as information from the transports. |
3 | SUPPORT |
This is full tracing through the stub, including detailed traces of control blocks, message information, etc. |
If you start your application with this environment variable set, a log file is created in the directory where your application is started. The name of the log file is pid.etb
csh or tcsh users use:
unsetenv ETB_STUBLOG
bsh, ksh or bash users use:
unset ETB_STUBLOG
Example of ETB_TRANSPORT
:
Shell | set the environment variable: | delete the environment variable: |
---|---|---|
C Shell | setenv ETB_TRANSPORT value |
unsetenv ETB_TRANSPORT |
Bourne or Korn Shell | ETB_TRANSPORT=value |
unset ETB_TRANSPORT |
EntireX Broker supports clustering in a high-availability scenario, using the environment variable ETB_SOCKETPOOL
. See Environment Variables in EntireX. This section covers the following topics:
See also High Availability in EntireX.
A TCP/IP connection established between stub and broker is not exclusively assigned to a particular thread. With multithreaded applications, two or more threads may use the same connection. On the other hand, if a connection is busy, another new one is created to exchange data.
In order to access the same broker instance in a clustering environment, an affinity between application thread and TCP/IP connection is needed to always use the same connection within an application thread. Therefore, an environment variable is evaluated to control the handling of TCP/IP connections.
If environment variable ETB_SOCKETPOOL
is set to "OFF" (ETB_SOCKETPOOL=OFF
),
an affinity between threads and TCP/IP connections is established. All
requests to one particular broker will use the same TCP/IP connection.
ETB_SOCKETPOOL
controls all TCP/IP connections.
Stubs ARFETB
and NATETBZ
always establish an affinity between subtask and TCP/IP connection.
Broker attribute CONNECTION-NONACT
is used by the broker to close TCP/IP
connections after the elapsed non-activity time. Omit this attribute
to keep the TCP/IP connection alive.
ETB_SOCKETPOOL=ON
is the default setting. In this case, an established
broker connection can be used by any thread if the connection is not busy.
Stubs with enabled socket pool (see environment variable ETB_SOCKETPOOL
)
can be configured to limit the size of the socket pool
(environment variable ETB_POOLSIZE
)
and to define the maximum wait time for a free connection
(environment variable ETB_POOLTIMEOUT
).