It's all about the answers!

Ask a question

What are 'JenaQueryService' jobs, why they are launched 30 minutes apart and why they kill our JTS (and RRC)?

Walter Cigana (1245) | asked Mar 07 '14, 11:29 a.m.
We are on CLM 4.0.5 and we have been seeing issues in RRC for a while now.

We can see that there are JenaQueryService jobs (user rrs-user) launched 30 minutes apart, and they run for 3h30 minutes each on average before finally exiting.  To add to that, we see some other jobs also launched as RRC users are working with the tool.

The issue from a user perspective is that this slows down loading views, or just makes them timeout.  We either see RRC giving up, or complaining that it cannot reach the server.  Granted we have big project areas (30,000+ artifacts), but I don't think this is what is the real issue.

From a monitoring point of view, this causes CPU usage to go up, even when relatively nobody is logged into the system.  Eventually, the jobs die and go away, or if that doesn't happen because people are using RRC, we need to restart the JTS, which affects everybody (RTC, RQM and RRC).  There is no 'nice and clean' way to kill the jobs, so we end up restarting the JTS.

So, can anyone tell me - partial answers are welcome :)
  1. What JenaQueryService jobs are?
  2. What is the rrs-user doing with this?
  3. Why it starts automatically every 30 minutes? 
  4. If we can change the interval between runs?
  5. ... and why this kills our JTS, sending the CPU usage through the roof.

2 answers

permanent link
Donald Nong (14.5k414) | answered Mar 09 '14, 11:32 p.m.
What you saw is the inner working of JTS with its index. This technote talks a bit about it.
Without any further details, it is really hard to say that the Jena Query Service "killed" JTS. If you regularly get 100% CPU or hang issues, you'd better open a support ticket and get the investigation going.

Walter Cigana commented Mar 10 '14, 10:24 a.m.

We do have support helping us with this, but it's really starting to affect the users and so we are trying to understand what we're seeing to possibly make changes that will help correct this.  No luck so far.

We regularly see many JtsQueryService processes started 30 minutes apart, and they run for 3.5 hours each before going away. 
Can you offer any other insight into why the process starts every 30 minutes and if this can be modified.  If so, how?  Thanks!

permanent link
Walter Cigana (1245) | answered Mar 11 '14, 6:10 p.m.
Just for everyone's knowledge:
The 30 minutes interval we're seeing is due to the fact that RRC is trying to sync some data and this times out  due to an RRC socket timeout.  When the 30 minutes expires, the process starts again since the sync is still needed.
It is also my understanding that this is fixed.  Just thought I'd share this info in case it can help someone else. 

In our case there is an underlying issue still to be discovered.

Your answer

Register or to post your answer.