Experimental Physics and
Industrial Control System

On Jul 22, 2020, at 4:28 PM, Tom Cumming via Tech-talk <tech-talk at aps.anl.gov> wrote:

I've been tasked with reviewing how we're currently handling ioc logging. Currently, the console concentrators we're using (Perle, WTI) are handling logging.

iocLogServer immediately comes to mind, but I swear that I saw somewhere that it's no longer going to be included in epics. Is that the case? Will it continue to be supported?

iocLogServer also does not conform to our standard logging practices. That means I'd have to change it but I believe these would be changes the community would enjoy.

1) Log rotation on the date/time instead of file size.

2) Individual log files per-ioc.

I'd like some feedback from folks!!!

The IOCs really don’t need any special kind of support from a log-server, so it is probably much better to pick a standard non-EPICS log-server package and just use that. All they do is open a TCP connection to a specific port on the indicated host, through which they then send text (ending in a newline) whenever they have something to log. I would like EPICS to adopt the convention that continuation lines in a multi-line message should always start with a whitespace character as that makes it easy to collect complete log messages, but there are quite a few places inside Base which don’t follow that and of course many messages come from outside of Base anyway.

The APS currently plan to use Logstash to collect messages and we will probably have that push them to an ElasticSearch database with a Kibana front-end once we get around to setting everything up. You probably already have some standard logging tool that should be able to handle them just as well once configured. I really don’t think it’s worth trying to fix the existing log-server code, I’m sure you have better things you could be spending time on.

HTH,

- Andrew