Start up Issues
As with any EnShell use, if there is a problem launching EnSight, be sure
to see if the EnShell network is starting correctly before invoking EnSight. Typically, the
person who set up EnShell for a site can determine this. If a EnShell network is running, use the
command enshell31 -cmd show_net
to display the running network, hosts
running the EnShells, and the various EnShell roles assigned to each EnShell. This command is
executed on the same computer running the EnShell that has the '-app' option. See the EnShell
documentation for further details.
In terms of RLSOS specifically, one EnShell must have the role RLSOS. One or more EnShells one level below this EnShell must have the role SOS. Finally, EnShells below these EnShells must have the role SOS_SERVERS (see Figure 1).
While EnSight Enterprise (formerly HPC+) may be used with RLSOS, it still has limitations affecting how the DRClients and CollabHub connect back to the Master Client and RLSOS. While EnShell is used to launch the DRClients and CollabHub, the DRClients and CollabHub do NOT use EnShell for establishing network communication to the Master Client nor RLSOS. This can lead to problems especially if firewalls or restrictive routers are used or when hosts have multiple network interfaces with varying hostnames. Ansys Support can advise.
Output
Due to the potential of running thousands to tens of thousands of EnSight
Servers, output from the Servers (or DRClients) can be overwhelming. More importantly, this
magnitude of output can actually throttle overall performance. To remedy this problem, EnShell
was enhanced for use with EnSight RLSOS to address this issue. EnShell adds a command line
option -app_output_buffering true|false
that controls output buffering
for any processes started by the EnShell.
If EnShell output buffering is turned on, output from its child processes (for example, the EnSight Server) is buffered until the process exits or until an error is detected in the output. Should an error be detected in the output, the buffer will be released for display. Furthermore, once the buffer is release, no additional buffering occurs until the process terminates.
The example EnShell network configurations provided with
ceistart
use this option to provide output buffering in a reasonable way along with
three selectable levels. You can choose to turn off all output buffering; this means that all
output from EnSight processes is displayed. Turning on output buffering means that no output
from EnSight processes is displayed (except for error conditions). The third option 'default'
shows all output from the EnSight RLSOS and CollabHub processes but buffers the output from the
Servers, DRClients, and child SOS processes. This option tends to give reasonable feedback
without adversely affecting performance. Turning output buffering 'off' may be useful for
debugging purposes but you should be mindful that the amount of output can be exceedingly large.
Also note that -app_output_buffering
option is independent from the -v
#
option which affects the amount of verbosity from the EnSight components (Client,
Server, SOS, etc.). For maximum output, use -app_output_buffering false -v
3
. For minimum output, use -app_output_buffering true -v
0
.
For any questions or problems with using EnShell and/or RLSOS, contact
Ansys Support (<support@ansys.com>
).