Hung Threads - Lawson crashes daily during business hours

We had this happen to us... Check to make sure that all of your Element Groups transferred/upgraded appropriately. The tools (lsdump/lsload) back when we did our upgrade did not transfer these across. *May have been fixed since then.

I believe I can help you with that, but I would need to look at your SystemOut.log and accesslogs.  If interested , send those to me at cadixon@corvondix.com.

Thank you Greg.  I have confirmed the Element Groups exist.

We've been dealing with this on/off since we moved to LSF 9.0.1 almost 2 years ago.  Currently we are on LSF 9.0.1.8.  In our most recent case with Lawson on this topic, they had me apply the tech patch for JT-336259 (that was early August 2012).  The system was stable for a while, but it's recently started hanging again occasionally although we haven't noticed the hung threads in the log files anymore.
One other thing I should mention is sometimes when our system seems "slow", simply stopping/restarting lase seems to help vs completely restarting Websphere.  That doesn't always work, but sometimes it does.
Good luck... I'd be interested to know if you ever get a lasting resolution.  We still have not.

We had similiar issues where lase was the culprit. This became much better once we vertically scaled (3 JVMs) and set min / max JVM performance parameters as recommended by Lawson Prof. Services.  Heavy Lawson Security activity seemed to be the issue for us. 

We have also scaled vertically to 3 App servers (3 JVMs). Also, if you are finding issues with lase and are using TDS, this link helped us quite a bit. Once we added indexes to some of the most heavily used searches, things straightened out.
http://www.ibm.com/develo.../library/t-tds-perf/

The JVM issue seems to be common. We too have mulitple JVM's. What I have seen talking to a number of customers is that they scaled to mulitple JVM's based on the needs of the business. So one hospital in our area scaled to accomodate Add-in calls, another LBI and we scaled our to accomodate MSCM calls. In any case, the scaling to mutliple app servers seems to have taken care of the issues. We also recycle websphere monthly during a standard downtime window that seems to help as well.

How many JVm's are you running currently?

Thanks everyone. We use TDS, are running 1 JVM, use LAUA security still. Lawson has provided a config change that has kept Lawson up and running longer (made it 2 weeks before another crash), but in the last handful of days we've experienced 2 more outages. Lawson's response to that was to install some logging programs they've provided to help pinpoint the root cause. Still no permanent solution, I'll keep you updated.

I understand that Lawson is recommending 2 or more JVMs on any new installations if you run Lawson Security. 1 JVM can sometimes struggle with a busy ERP system under Lawson Security.

When you asked about heavy Lawson Security activity, we have correlated issues to running larger security reports or making many changes in classes, especially when multiple users are in the security application (i.e. during our security implementation).

very similar issue about a few yrs back. i created additional cluster servers, and moved MSCM onto its WAS, as 2nd instance. since then, no issue for me.

Posted By Brian Allen on 09/25/2012 04:37 PM
We had similiar issues where lase was the culprit. This became much better once we vertically scaled (3 JVMs) and set min / max JVM performance parameters as recommended by Lawson Prof. Services.  Heavy Lawson Security activity seemed to be the issue for us. 

So what did they have you set the min/max to?

LPS had us set the following JVM settings.  These recommendations were from their discussions with Redmond / IBM that best works with how Lawson runs.  They recommended NOT going with the max settings due to how garbage collection works.

Sorry for the delayed response... too many projects lately.
min 384
max 1586

Generic JVM Arguments:
-Xgcpolicy:gencon

We've had no issue with Lawson hangs since late October, so my team and Lawson worked together to determine why/how it was fixed.
Memory to Memory replication was on... we turned it off (Lawson recommends this anyway). A new LADB and service programs were provided by Lawson as well. We created a lower environment, prepared it as we had Prod at go live and successfully replicated the hang (Excel spreadsheet designed to pump in transactions mimicking a custom process that creates AP invoices, AP vendors and AR comments real time). Then tried different versions of LADB and service programs that were made available and found that a particular version of the service programs allowed semaphores to open/close correctly and we are now ok.
Lawson is preparing a technology patch.