Hi John, Thanks for the fast response! On May 21, 9:09 pm, John Mettraux <[email protected]> wrote:
...*snip*... > > 1. A newly created process never gets expanded/processed. It is > > stuck in the "define" stage, with only 1 expression active. ie. the > > process creates but never goes anywhere. > > The way you describe it, it seems that all the processes never go further the > first step. But you mention previously "not always". That's weird. It is a random occurrence. I'd say maybe 90% of the time the processes behave. When they don't, this is one of the two symptoms. In this case it acts as if no worker is running (but there is one, afaik), and no process gets instantiated. > > > 2. Steps (expressions) that are completed, do not get removed from > > the storage. For example if I have a sequence with step A and step B, > > when step A completes my process will then show *both* A and B as > > active. > > Is this consistent with the two storage you've been trying? Yes. What seems to be happening here is that the step completes but the record in the documents table hangs around. This is a problem for me because I have a UI that queries the active steps of a process and displays them to the user. If I manually delete the record from the database, all seems to be well. > > > I have tried using both file system and sequel storage drivers. > > Now time for the questions: > > - which platform, OS? CentOS 5.5 on a x86 VPS > - which Ruby version? 1.9.2-p290 > - which version of ruote? (master is recommended) 2.2.0, the last official release : ) I suppose I can give the latest stable git version a try, if you think it will help. I'm one of those guys that has a preference for only using "official" releases in production. I'm not religious about it though. As long as I can configure my app to pull a specific git snapshot, and it works, that is good enough for me. I just don't like being surprised by upstream updates when I'm not expecting/ready for them. > - single worker architecture? I think this could well be the problem, but I'm not sure which is why I asked for help. To me, the problem smells like it might be locking, or a race condition. I have a single worker in each instance of the rails app. However, I'm running under Apache+Passenger and I have multiple (currently 6) passenger instances => multiple instances of rails app => multiple workers, one worker per instance of my rails app. If that makes sense. Short-term, scaling back to one worker may be an option. But longer term I think I will need multiple workers to get the throughput I need. Also (see below) I see this problem in dev sometimes with single worker, as well. > - sequel against which database? Is the database on the same host as the > workers? MySQL 5.1. Same host as everything else; workers, apache, passenger, rails app. It did occur to me this could be a sizing problem. I'm starting to run out of things to try / places to look, so rather than guess I thought I'd ask the experts. I suppose I should add, these problems occur far less often in development, running under "rails server" (WEBrick) where I *know* there is only one worker. But they do occur. I get the problems if I start a lot of processes quickly, close together in time. Our process definition is still changing, and I have a script I run to re-launch all the processes with the new definition. If I have, say, 10-15+ definitions getting launched one after the other, I will sometimes see the above symptoms in dev as well. -- you received this message because you are subscribed to the "ruote users" group. to post : send email to [email protected] to unsubscribe : send email to [email protected] more options : http://groups.google.com/group/openwferu-users?hl=en
