Experimental Physics and
Industrial Control System

Ralph Lange <[email protected]> · Fri, 31 Oct 1997 11:38:36 +0100 (MET)

Just a comment/question on a detail - sorry if this sounds too obvious ...

> [... becon time / reboot ...]
> the reconnect problem for IOCs that take minutes to boot (we're still 
> on beta4).

I don't see why the beta4-to-beta11 step would increase an IOCs boot up
speed - do your IOCs have NFS access to their boot server?

We once had a problem where one of the "larger" IOCs took about 10 minutes
to boot. It turned out that we forgot to have this IOC listed in the
/etc/exports file of the boot host so that vxWorks used the fallback method
(remsh) to load the database. If you use templates, the .db file gets
opened once per instantiation (this leads to an overhead of ~1 second per
instance!!).

We're also running beta4 on all our crates. The statistics for our largest
IOC are

CPU:      MV162-042 (20 MHz, 8 MB)
OS:       vxWorks 5.2
EPICS:    3.13.0.beta4
Database: 510 template instances create 1726 soft and 514 hardware records
          (sums up to 2240 records; 6.5 MB of allocated memory)
Seq:      3 state sets
Load:     ~50% (when idle)

Boot up:  reboot until iocInit  1:25 min
          iocInit               0:55 min

(Note: iocInit takes long because it has to set up two intelligent field
       bus cards where each data object has to be configured separately.)

(BTW: This was the IOC with the 10 min boot up time using remsh access.)

Hope this helps (or doesn't hurt, at least) ...

Ralph

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System