Experimental Physics and
Industrial Control System

"Brian McAllister" <[email protected]> · Tue, 06 Nov 2001 17:36:10 -0500

Folks -

I was reviewing this thread and I find myself with a question.

Are those of you using "console logging" (through xterms or other) using
this instead of the EPICS iocLogServer ?

Or in addition to it ?

Here at Bates, we have serial ports on Linux machines placed around the
facility connected to the IOC consoles (which also allows them to be used
as local OPIs).  They are accessed using Kermit.  We do not leave these
connections open for logging purposes.  They are used to verify boot
behavior and for debugging purposes.

For logging, we run a copy of iocLogServer for each IOC.  One server could
be used instead, but then the output needs to be extracted before output
from any single system can be examined.  Also, since the log files have a
maximum size limit, after which they are wrapped and overwritten from the
top, an IOC that starts spewing errors can cause the loss of information
from the others.

All the servers run on the same system, using different port numbers.  At
present, we only have 15 IOCs, and while I expect that to grow some, I
don't anticipate issues with the number of logging processes. For some of
you, that may not be the case.

I am not aware of any important EPICS messages that are output only to the
console, rather than through the error logging facilities.

Other than possibly missing something coming from VxWorks itself, is there
any reason why direct logging of the console is necessary ?

----
Brian McAllister                    Controls Programmer/Beam Physicist
[email protected]                        MIT-Bates Linear Accelerator
(617) 253-9537                                           Middleton, MA

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System