pgsql: Enable Unix-domain sockets support on Windows

On Sat, Mar 28, 2020 at 7:37 PM Peter Eisentraut <peter@eisentraut.org> wrote:

Enable Unix-domain sockets support on Windows

+
+/*
+ * Windows headers don't define this structure, but you can define it yourself
+ * to use the functionality.
+ */
+struct sockaddr_un
+{
+   unsigned short sun_family;
+   char sun_path[108];
+};

I was going through this feature and reading about Windows support for
it. I came across a few links which suggest that this structure is
defined in <afunix.h>. Is there a reason for not using this via
afunix.h?

[1]: https://devblogs.microsoft.com/commandline/af_unix-comes-to-windows/
[2]: https://gist.github.com/NZSmartie/079d8f894ee94f3035306cb23d49addc

--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com

On 2020-06-26 14:21, Amit Kapila wrote:

On Sat, Mar 28, 2020 at 7:37 PM Peter Eisentraut <peter@eisentraut.org> wrote:

Enable Unix-domain sockets support on Windows

+
+/*
+ * Windows headers don't define this structure, but you can define it yourself
+ * to use the functionality.
+ */
+struct sockaddr_un
+{
+   unsigned short sun_family;
+   char sun_path[108];
+};

I was going through this feature and reading about Windows support for
it. I came across a few links which suggest that this structure is
defined in <afunix.h>. Is there a reason for not using this via
afunix.h?

[1] - https://devblogs.microsoft.com/commandline/af_unix-comes-to-windows/
[2] - https://gist.github.com/NZSmartie/079d8f894ee94f3035306cb23d49addc

If we did it that way we'd have to write some kind of configuration-time
check for the MSVC build, since not all Windows versions have that
header. Also, not all versions of MinGW have that header (possibly
none). So the current implementation is probably the most practical
compromise.

--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

On Sat, Jun 27, 2020 at 3:06 PM Peter Eisentraut
<peter.eisentraut@2ndquadrant.com> wrote:

On 2020-06-26 14:21, Amit Kapila wrote:

On Sat, Mar 28, 2020 at 7:37 PM Peter Eisentraut <peter@eisentraut.org> wrote:

Enable Unix-domain sockets support on Windows

+
+/*
+ * Windows headers don't define this structure, but you can define it yourself
+ * to use the functionality.
+ */
+struct sockaddr_un
+{
+   unsigned short sun_family;
+   char sun_path[108];
+};

I was going through this feature and reading about Windows support for
it. I came across a few links which suggest that this structure is
defined in <afunix.h>. Is there a reason for not using this via
afunix.h?

[1] - https://devblogs.microsoft.com/commandline/af_unix-comes-to-windows/
[2] - https://gist.github.com/NZSmartie/079d8f894ee94f3035306cb23d49addc

If we did it that way we'd have to write some kind of configuration-time
check for the MSVC build, since not all Windows versions have that
header. Also, not all versions of MinGW have that header (possibly
none). So the current implementation is probably the most practical
compromise.

Fair enough, but what should be the behavior in the Windows versions
(<10) where Unix-domain sockets are not supported? BTW, in which
format the path needs to be specified for unix_socket_directories? I
tried with '/c/tmp', 'c:/tmp', 'tmp' but nothing seems to be working,
it gives me errors like: "could not create lock file
"/c/tmp/.s.PGSQL.5432.lock": No such file or directory" on server
start. I am trying this on Win7 just to check what is the behavior of
this feature on it.

--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com

On 2020-06-27 13:57, Amit Kapila wrote:

Fair enough, but what should be the behavior in the Windows versions
(<10) where Unix-domain sockets are not supported?

You get an error about an unsupported address family, similar to trying
to use IPv6 on a system that doesn't support it.

BTW, in which
format the path needs to be specified for unix_socket_directories? I
tried with '/c/tmp', 'c:/tmp', 'tmp' but nothing seems to be working,
it gives me errors like: "could not create lock file
"/c/tmp/.s.PGSQL.5432.lock": No such file or directory" on server
start. I am trying this on Win7 just to check what is the behavior of
this feature on it.

Hmm, the only thing I remember about this now is that you need to use
native Windows paths, meaning you can't just use /tmp under MSYS, but it
needs to be something like C:\something. But the error you have there
is not even about the socket file but about the lock file, which is a
normal file, so if that goes wrong, it might be an unrelated problem.

--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

On Sun, Jun 28, 2020 at 2:03 PM Peter Eisentraut
<peter.eisentraut@2ndquadrant.com> wrote:

On 2020-06-27 13:57, Amit Kapila wrote:

BTW, in which
format the path needs to be specified for unix_socket_directories? I
tried with '/c/tmp', 'c:/tmp', 'tmp' but nothing seems to be working,
it gives me errors like: "could not create lock file
"/c/tmp/.s.PGSQL.5432.lock": No such file or directory" on server
start. I am trying this on Win7 just to check what is the behavior of
this feature on it.

Hmm, the only thing I remember about this now is that you need to use
native Windows paths, meaning you can't just use /tmp under MSYS, but it
needs to be something like C:\something.

I have tried it by giving something like that.
After giving path as unix_socket_directories = 'C:\\akapila', I get
below errors on server start:
2020-06-29 08:19:13.174 IST [4460] LOG: could not create Unix socket
for address "C:/akapila/.s.PGSQL.5432": An address incompatible with
the request
ed protocol was used.
2020-06-29 08:19:13.205 IST [4460] WARNING: could not create
Unix-domain socket in directory "C:/akapila"
2020-06-29 08:19:13.205 IST [4460] FATAL: could not create any
Unix-domain sockets
2020-06-29 08:19:13.221 IST [4460] LOG: database system is shut down

After giving path as unix_socket_directories = 'C:\akapila', I get
below errors on server start:
2020-06-29 08:24:11.861 IST [4808] FATAL: could not create lock file
"C:akapila/.s.PGSQL.5432.lock": No such file or directory
2020-06-29 08:24:11.877 IST [4808] LOG: database system is shut down

But the error you have there
is not even about the socket file but about the lock file, which is a
normal file, so if that goes wrong, it might be an unrelated problem.

Yeah, but as I am trying this on Win7 machine, I was expecting an
error similar to what you were saying: "unsupported address family
...". It seems this error occurred after passing that phase. I am not
sure what is going on here and maybe it is not important as well
because we don't support this feature on Win7 but probably an
appropriate error message would have been good.

--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com

On 6/28/20 4:33 AM, Peter Eisentraut wrote:

On 2020-06-27 13:57, Amit Kapila wrote:

Fair enough, but what should be the behavior in the Windows versions
(<10) where Unix-domain sockets are not supported?

You get an error about an unsupported address family, similar to
trying to use IPv6 on a system that doesn't support it.

BTW, in which
format the path needs to be specified for unix_socket_directories?  I
tried with '/c/tmp', 'c:/tmp', 'tmp' but nothing seems to be working,
it gives me errors like: "could not create lock file
"/c/tmp/.s.PGSQL.5432.lock": No such file or directory" on server
start.  I am trying this on Win7 just to check what is the behavior of
this feature on it.

Hmm, the only thing I remember about this now is that you need to use
native Windows paths, meaning you can't just use /tmp under MSYS, but
it needs to be something like C:\something.  But the error you have
there is not even about the socket file but about the lock file, which
is a normal file, so if that goes wrong, it might be an unrelated
problem.

It needs to be a path from the Windows POV, not an Msys virtualized
path. So c:/tmp or just /tmp should work, but /c/tmp or similar probably
will not. The directory needs to exist. I just checked that this is
working, both in postgresql.conf and on the psql command line.

cheers

andrew

--
Andrew Dunstan https://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

On Mon, Jun 29, 2020 at 8:48 PM Andrew Dunstan
<andrew.dunstan@2ndquadrant.com> wrote:

On 6/28/20 4:33 AM, Peter Eisentraut wrote:

On 2020-06-27 13:57, Amit Kapila wrote:

Fair enough, but what should be the behavior in the Windows versions
(<10) where Unix-domain sockets are not supported?

You get an error about an unsupported address family, similar to
trying to use IPv6 on a system that doesn't support it.

BTW, in which
format the path needs to be specified for unix_socket_directories? I
tried with '/c/tmp', 'c:/tmp', 'tmp' but nothing seems to be working,
it gives me errors like: "could not create lock file
"/c/tmp/.s.PGSQL.5432.lock": No such file or directory" on server
start. I am trying this on Win7 just to check what is the behavior of
this feature on it.

Hmm, the only thing I remember about this now is that you need to use
native Windows paths, meaning you can't just use /tmp under MSYS, but
it needs to be something like C:\something. But the error you have
there is not even about the socket file but about the lock file, which
is a normal file, so if that goes wrong, it might be an unrelated
problem.

It needs to be a path from the Windows POV, not an Msys virtualized
path. So c:/tmp or just /tmp should work, but /c/tmp or similar probably
will not. The directory needs to exist. I just checked that this is
working, both in postgresql.conf and on the psql command line.

Okay, thanks for the verification. I was trying to see its behavior
on Win7 or a similar environment where this feature is not supported
to see if we get the appropriate error message. If by any chance, you
have access to such an environment, then it might be worth trying on
such an environment once.

--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com

On 6/30/20 12:13 AM, Amit Kapila wrote:

On Mon, Jun 29, 2020 at 8:48 PM Andrew Dunstan
<andrew.dunstan@2ndquadrant.com> wrote:

It needs to be a path from the Windows POV, not an Msys virtualized
path. So c:/tmp or just /tmp should work, but /c/tmp or similar probably
will not. The directory needs to exist. I just checked that this is
working, both in postgresql.conf and on the psql command line.

Okay, thanks for the verification. I was trying to see its behavior
on Win7 or a similar environment where this feature is not supported
to see if we get the appropriate error message. If by any chance, you
have access to such an environment, then it might be worth trying on
such an environment once.

I haven't had a working Windows 7 environment for quite some years.

cheers

andrew

--
Andrew Dunstan https://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services