SOAP Errors in PDMLink with Wildfire 5

[pstephens]

Hello All,

Not sure if this is exactly the right place to post, but here goes anyway.

We are currently testing PDMLink 9.1 M050 with Pro/E WF5 M060 with the intention of upgrading in a couple of months.

We are set up as a replica site in Australia with our own local content cache server and the master server is in Seattle.

Other replica sites in the US are also experiencing similar issues, but not as frequently as us.

I am testing using the embedded browser in Pro/E and am getting many (more than 6 or 7) SOAP errors per day along with the "The server is currently unavailable.Do you want to switch to offline mode?"

The SOAP errors can happen just by launching WF5 and trying to go to server manager, or import to workspace, add to workspace, change a window configuration.

PDMLink and Pro/E would lock for over 5 minutes then return the error messages.

I raised a call with PTC and they suggested some changes which I have implemented and they helped a bit in that the Pro/E lock ups were not as severe, but am still getting the disconnection and SOAP messages.

We set the logging options and the environment variables as per option 2 in TPI 141793 as well as the Client Property Settings for our config.pro from the PTC Tuning Recommendations document.

PDMLink and Pro/E are definitely running better but I am still frequently getting the following errors:

SOAP Client Error(line1, column 73655):An exception occurred! Type:UnexpectedEOFException, Message:The end of input was not expected.

SOAP Client Error(line1, column 49004):The input ended before all started tags were ended.Last tag started was 'value'.

SOAP Client Error: Failed to open SOAP RPC call response stream.

SOAP Client Error: Failed to finish writing SOAP call request.

Any suggestions would be greatly appreciated, we are currently just treading water on this issue.



TIA


Paul Stephens

CAD Systems Supervisor

Kenworth Trucks Australia

[DanielReid]

To add to Paul's comments (I'm the admin in Seattle), I am starting to
tweak the Apache KeepAliveTimeout, starting with 15 seconds instead of
the OOTB 5 seconds. Has anyone else had to adjust their Apache timeouts
or something similar?

I also read that proxies can cause issues but have not gotten into
details to talk with our IT folks about that. At least I was finally
able to dismiss much of the shotgun spray of possible solutions PTC
threw at us as too old or not applicable to our system.

Also, even "local" users (1ms-2ms latency, 100Mb connections) are
experiencing SOAP errors on occasion so while increased latency does
seem to mean increased frequency, it isn't only a problem for distant
users.

Thanks!

[ArturPonurski]

We have SOAP errors with some users at various global locations. We
started on Apache KeepAlive 5 then 15 then 30 with finally 60 seconds.
Yes, some users are better now but some still getting SOAP errors.



We are looking at all possible settings starting at: client machines (IE
7 settings), WAN accelerator(riverbed), firewall, load balancer (F5),
pocket shaper (Blue Coat), windchill server - it is quite a task to
review all of them and make sure all are aligned! ......



I will monitor this thread very closely and I will try to report on any
findings on my side.





Our WC configuration: cluster with 3 Method servers on slave and one
Background Method server on master

Windchill 9.1 M50 on RHEL 5.5

Apache 2.2.14 (Unix)

Tomcat 5

JDK 1.6.0_18

[DanielReid]

To follow-up, I ended up stopping at a KeepAliveTimeout of 45 seconds.
There was improvement for Australia right away at 15 but still some
users at other sites with problems. I actually still have a couple
users with problems in Texas but I think there might be some localized
network issues or local config problems there.

One thing I found I have to watch for now is to be sure the MaxClients
is high enough, since I could end up with more active connections
requested than available. There's some good information on
KeepAliveTimeout in the Apache documentation and in an older but good
blog here:
.
I have enough RAM on my server to allow for increasing that.