Experimental Physics and
Industrial Control System

Mark Rivers <[email protected]> · Mon, 12 Nov 2012 17:52:17 +0000

Hi Ralph,

> This option never existed in the official procServ sources.

Whoops, you're right, that was added in a local modification by Matt Newville that I was not aware of.

>> - The timestamp output is totally mixed in with the output, making it very difficult to read (see below).

> I will look into this.
> Which platform are you running procServ on?

Linux, Fedora Core 15, gcc 4.6.3.

Thanks,
Mark

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Ralph Lange
Sent: Monday, November 12, 2012 10:59 AM
To: EPICS Tech Talk
Subject: Re: procServ softIOC server - V2.6.0 released

Hi Mark,

On Mon Nov 12 2012 17:14:03 GMT+0100 (CET), Mark Rivers
<[email protected]> wrote:
> I installed procServ 2.6.0 because I really want to use the timestamp feature on the log files.  One important application for us is using procServ to run the "telnet" command to our terminal servers connected to the VME console port on vxWorks IOCs.  This allows multiple users to connect to the VME crate, and produces nice log files.  That vxWorks output previously had no timestamps, so it was hard to know when errors occurred.

For this situation, conserver [1] is a *much* better suited application.
(Which also does nice timestamps.)
But, however, procServ should be able to handle it.

> However, procServ 2.6.0 seems to have a number of problems.
>
> - The command to exit procServ no longer appears to be configurable.  Previously it could be configured with -Q.  Now it appears to be hardcoded to be ^Q.  However, ^Q is needed when logged into a vxWorks system for XON/XOFF control (type ^S to suspend output, ^Q to resume).  We were previously setting -Q to ^E.

This option never existed in the official procServ sources.
Also, the ^Q (quit) character is only caught and interpreted when the
child process is down, i.e. when you are not connected to the remote
vxWorks system. While the child is alive and you are connected, ^Q will
be forwarded to the child process.
Which means the -Q option, that never existed, does not seem necessary
in the first place.

> - The timestamp output is totally mixed in with the output, making it very difficult to read (see below).

I will look into this.
Which platform are you running procServ on?

> - The log file has ^M characters in it at the end of each line.  This is not a new problem, 2.4.0 also had this problem.

The ^M characters look like being part of a telnet protocol.
I suspect this is an issue with the connection between the telnet client
(that you start as procServ child) and the terminal server on the other end.
Which would place it outside the scope of procServ.

Can you test with a simple non-telnet client (e.g. a shell)?

Thanks,
~Ralph

[1] http://www.conserver.com

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System