AnsweredAssumed Answered

Service type events not firing

Question asked by diogo.nobrega on Sep 14, 2016
Latest reply on Sep 21, 2016 by diogo.nobrega

Hi everyone,


We have been experiencing an issue regarding service type events in one of our Service Desk instances. The events in call_reqs are not firing when they are reached. As a result, the events remain in "Pending" state forever (unless the request is Canceled or Closed, when the event is canceled too).


This is weird, since it was all working since two years ago, and suddenly we got this problem. I can reproduce this in a replicated environment, and it looks like this issue is happening on all requests/incidents and service types.


Because we are using SDM 12.7 which is out of service, I was investigating this on my own. I have discovered that every time an event is supposed to be fired, this error is written on stdlog:


09/14 16:23:19.14 VCSPECONSRV122 animator_nxd 5468 ERROR       animator_nxd.c 1839 10 in animator ANI:402325 (cr:652896) for atev:627039 in method animation_complete
09/14 16:23:19.29 VCSPECONSRV122 animator_nxd 5468 SIGNIFICANT animator_nxd.c 2121 Deleted animator ANI:402325 (cr:652896) for atev:627039 due to earlier failures during firing


Now, some theories about what could be happening: "10" may be an error code (or else, an "dead mark") for that event/animator. In "ani" table, most rows have the "10" value in "a_delta" column. But even the rows with "0" value were eventually "killed" too, which is strange.


Maybe there is a flag or a black list containing all events that are "died" for some reason (probably the "earlier failures during firing). When an event is triggered, maybe this animaton_complete function is called, and then it checks if the event is on "black list" or not.


So this raise some questions:

1. What this animation_complete method really does?

2. Why are these events are being "marked" as "failed"? And where?

3. How can I "clean" these events so the animator can fire again?


Could anyone help us on this?


PS: we're using classic SLA option since the beginning.