Kbase 13968: What the .ld, .l2, .l2, ... unix socket files are about
| Autor |
Progress Software Corporation - Progress |
| Acesso |
Público |
| Publicação |
5/10/1998 |
|
What the .ld, .l2, .l2, ... unix socket files are about
The following is an explanation of what the dbname.ld,
dbname.l1, dbname.l2, etc... files are used for, which
you may see created by the broker and servers under unix.
These files are sockets. They are there to support a
communication protocol called "Unix domain sockets". The
Berkeley socket interface allows use of a number of
communication protocols. The most commonly used one is
TCP/IP, but there are many others, such as UDP and Unix
domain sockets. TCP/IP is part of a family of over 100
related protocols which are used for many different
purposes (NFS, e-mail, telnet, rlogin, etc.).
Unix domain sockets is a feature of some Unix systems that
allows processes on the *same* machine to communicate
through the Berkeley socket interface. It is very similar
to named pipes. In fact, some Unix systems use this
mechanism to implement support for named pipes internally.
Addresses for Unix domain sockets are determined through
pathnames in the filesystem. This is why the servers and
broker create these "files". They are created in case a
client wants to connect through this protocol.
The "-cl" connection
option allowed a client to use this feature. It never
worked very effectively and no one used it. A severe
drawback of our product's use of this feature is that there
can be only 50 server processes for each database. This is
because each of the socket files must have a unique name
and the algorithm we use for inventing names limits us to 50.
We no longer support this feature and it has been removed
in one of the version 7 releases, so these 'files" are no
longer created. The limit on 50 servers has also been
removed. The 7.3A doc says the limit is now 512 for max
servers.
Progress Software Technical Support Note # 13968