Experimental Physics and
Industrial Control System

"Kasemir, Kay" <[email protected]> · Fri, 07 Jan 2011 16:49:07 -0500

Hi:

OK, that somewhat explains it.

You happen to have a value of 0.000/INVALID with a time stamp of 1989/12/31
19:00, i.e. a raw EPICS time stamp of zero, at the start of your live data.

The CSS Data Browser prefers 'live' data. It will request archived data for
the time range of the plot, but if you have live data from 1989/12/31 on, it
will only use the live data, not show archived data for that time period,
because it assume live data is the best possible quality.

The Archive Viewer "worked" because it only uses archived data, no live
data.

Not related to this at all, but since you asked why the Data Browser pings
the old Channel Archiver network data server twice: The protocol was such
that the data server would send the requested data, but not more than I
think 10000 samples in one batch. The client then has to ask again. The way
it was designed, the client has no idea if the received data is complete, or
if there's more. So it asks more for data from the end of the last result on
until there simply is no more data returned.

But back to your original problem: The big question is why you have an
initial live value with time stamp zero. Is that really the time stamp of
the PV on the IOC? When you tell the Data Browser to use the sample period
0, it uses a CA monitor, and you get the original 0 time stamp from your IOC
-> That would explain your problem. When you use the sampled mode where the
Data Browser periodically saves the most recent 'live' sample, it uses the
host clock.

Thanks,
Kay

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System