cancel
Showing results for 
Search instead for 
Did you mean: 

Portal running on VMware/win2k3/MSSQL that goes down with 4 users logged on

Former Member
0 Kudos

Hi

We have a portal running on VMware/win2k3/MSSQL that goes down with 4 users logged on.

We get a "SAPWebAS Engine is starting.......Dispatcher running but no server connected" error with very few users doing regular work.

At first it happened with only 1 user logged on. After this we boosted RAM to 16 GB (SAP says 4 GB is enough for 100 users). After this we still get the error, but the users can work for some time.

Its a bit strange, I cannot see anything regarding this in the logs/dev traces. However there is entries in the event viewer. Event Id: 35712. Description: SAP-Basis System: Operating system call recv failed (error no. 10054). After this we manually have to restart the portal.

We suspect it might be the VMware or Java VM, but is in doubt wich way we should direct the further investigation. Have any experienced similar and what was the problem then?

Regards

Kim

Accepted Solutions (0)

Answers (4)

Answers (4)

Former Member
0 Kudos

Hi,

Check the jcontrol log in work directory. If it says license check failed, go to VA(once u restart it will stay for abt 30 min.) check whether you have valid licenses.

You need portal_ora as well as Netweaver_ora.

VA-server-services-licensing adapter.

Apply the license using file upload.

Former Member
0 Kudos

HI,

Does your portal restart at regular interval? If it is so it may be problem with your saplicense.

Please check whether your Netweaver License is properly installed or not. Installing license again should solve your problem.

Jay

Former Member
0 Kudos

No, its not on a regular interval. I have checked license (dual stack) and its ok.

TomCenens
Active Contributor
0 Kudos

"After this we boosted RAM to 16 GB"

I hope you made additional nodes ? Because if you only have one server node with 16 GB assigned it will not work either.

2 GB heap size maximum per node should be assigned and 2 GB should be enough assigned to the J2EE engine itself to have a bunch of users work on it without problems. Create additional nodes if you require more resources.

Former Member
0 Kudos

We have tryed using more nodes, but then it goes down even faster. So we are back to single node, with 2GB heap size (which should be enough for 300 users).

Former Member
0 Kudos

In dev_server0 log there was a non-zero exit code, which I send to SAP in a oss message yesterday afternoon (CET). However I continue work in parallel with this.

There was nothing in the rest of the dev_traces. No indication of memory problems, which compared with there is a bug in VMware when using AMD processors makes us suspect VMware. When using AMD and VMware you can only assign one virtual CPU.

I will though check the params again.