DataStage comes up before LDAP, userIDs are invalid

m1kebaker · Post by **m1kebaker** » Fri Mar 30, 2012 1:00 pm

We run on AIX servers.
When we have an automated server restart, upon startup DataStage obviously validates user IDs before the LDAP server has made those user IDs available. DataStage comes up but is in a crippled condition because it fails to acknowledge our LDAP user IDs. The only working user ID is dsadm which is local. This does not however stop scheduled jobs from running (correctly). To fix the LDAP ID problem, it is necessary to bounce DS. Depending upon time of day and job scheduling, it can be 6-8 hours before we can get a clean window to bounce DS. Meanwhile, staff with LDAP ID access is unable to access the environment.
Secondly, how can we be sure what condition the environment is in if it went down unexpectedly? It seems that we desire a human to evaluate the environment and to return DataStage to service in a managed manner. We, like everybody else, have those few processes that if interrupted must have a specific and critical recovery procedure.
We also monitor each DS environment by periodically connecting from other servers and when one is unreachable, the alarms start going. When DataStage goes down, as long as one of our other servers is still up, we will know about it.
It seems to me that we very much do desire to manage our DataStage restarts and not just have them autostart regardless of what was occurring when it went down!

chulett · Post by **chulett** » Fri Mar 30, 2012 2:55 pm

Make sure you realize that you are replying to a six year old post discussing the 7.x architecture - the new world bears very little resemblance to the old world. Kind of a Christopher Columbus expression but still...

Let's get you into your own thread in the proper forum.

Edited to add: split from this post.

qt_ky · Post by **qt_ky** » Fri Mar 30, 2012 6:05 pm

I've run DataStage on AIX and Solaris servers without restarting the servers or having them restart unexpectedly for years at a time. I've never seen the DataStage Engine go down unexpectedly. I guess I've been lucky. My first concern here would be to address why the server would ever restart unexpectedly.

Your scheduled jobs still do run correctly? Are they all scheduled under the dsadm ID?

Can you determine how much startup delay is needed before the LDAP connectivity is ready after a server restart? Add a sleep <seconds> command into the DataStage startup script according to what LDAP requires.

If your startup script includes the node agents and metadata server plus the DS Engine then you may want to add a few sleep commands between each step anyway. There are some internal checks going on at startup and if your server is crashing while lots of jobs were running, there can be small delays (60 seconds).