EG 5.1 Loosing Work Library

[arnab61]

It has been recently observed that after migrating user from EG 4.3 to 5.1 the work libraries are getting lost.If a user opens a session and started writing to work library there is no problem till the time the user is accessing the work location too frequent.

After few hours if he has tried to access the datasets in work library the datasets are not there any more.

The backend investiation suggests that the Connection from SAS EG 5.1 is being reset and a new connection is created. The new connection is referring to a new location and the old location is existing with the datasets. But the seession has already lost the old work directory.

The possibility of Network Disrruption has already been tested and thre is no problem observed.

Please let me know if there is any known issue. Is there any way we can prevent it from happening?

[jakarman]

Review the parallel processing option with EGuide 5.1

There are many/several sessions going to run http://support.sas.com/resources/papers/proceedings13/138-2013.pdf (page 30) and can be configured.

http://support.sas.com/kb/51/225.html. When used every session will use his own new SAS-work and own macro vars etc. When set not to run parallel it  will update those settings in the session that serves as Eguide visualisation

The firewalls behavior is indeed another root cause. They will interrupt and break a client/server connection commonly a 2 hour interval is set.

This disconnection is based on measuring sessions and is intended behavior (regulations). That is something different as network disruptions.  

The requirement is that applications (SAS Oracle SSH etc better indication the middelware)  should send session-keep alive messages.

For SSh this is wellknown, with SAS its just coming in 9.4 SAS(R) 9.4 Intelligence Platform: Application Server Administration Guide

[SASKiwi]

As Jaap says, firewall rules are a possble cause of your problem as it is common practice these days to have firewalls around servers. In our SAS environment we have had to change the default firewall timeout to at least 8 hours.