Jobs behavingly strangely in 8.2.0 - may be related to XML file editing? - VisualCron - Forum

Community forum

rprastein
2016-11-17T18:10:33Z
Hi,

I have several jobs that I am working with, as I try out 8.2.0 (I'm a long-time user of 5.7.8). These jobs were working on my DEV server. They were working in VAL when I first deployed them there, by exporting all of the settings from DEV, making some edits to the InternalSettings.xml and some of the other xml files, and then importing to the VAL server.

I have since then done additional work on the DEV server, checked the settings XML files into revision control, and attempted to push selected changes to the VAL server from the revision control.

It is VERY difficult to compare the XML files, especially for the Jobs.xml file. The nodes aren't in the same order, and there is a lot of state information that I don't care about, intermixed with the job definition information. In order to compare, I had to sort the nodes by <id> and ignore a lot of fields relating to standard or error output, time last run, and the like.

After using this sorted compare and merge process, I find I have several jobs that no longer run properly. I suspect the problem is related to sorting the xml files, but I don't know for sure, because I've only recently installed 8.2.0 (I had 8.1.2 before, I think, when I first created some of these jobs).

I am seeing two problems:
1. On some jobs, a task (most often the first task) runs repeatedly, in an infinite loop, without progressing to the next task.
- stopping or de-activating the task does not stop this behavior; I have to stop the job and then de-activate it.
2. On some jobs (some of the same jobs as above, and some others), the a task (often the last task, but not always) appears to be hanging at 99%
- if I view the running processes, there is no process running
- the job itself shows a status of waiting
- if I disconnect and reconnect the client, the task shows as waiting rather than as 99%, so this appears to be a GUI error

What can be causing this?

My first thought was the sorting of the settings files might be the problem, but one job that is sorted the same in DEV and VAL has a different task affected with the infinite loop repeating problem depending on the server, but they are sorted identically.

As a side note, it would be helpful to know which elements in which of the settings files can safely be re-ordered, and which cannot. For example, it is apparent that the subelements of the TTime elements, months, days, intervaldays, intervalweekdays, hours, minutes, and seconds, the document order matters (the subelements all have a name of <boolean>).

Thanks,

Rebeccah
rprastein
2016-11-17T19:12:19Z
Work-around for task repeating in infinite loop: Stop job, open job for editing, save job (with out editing).
Sometimes this results in the repeating task now showing (and stuck at) 99% complete, sometimes the repeating task now shows waiting.
Work-around for task showing stuck at 99%: disconnect and reconnect the client to the server.

After that, all of the jobs and tasks appear to be behaving properly.
What is causing this? Is it a function of changing job settings while a task is in the process of running? I've looked at my reformatted XML files, and I really can't see anything there that should cause any problem.

Thanks,

Rebeccah
Support
2016-11-18T10:04:46Z
Why do you edit the XML files? Are you only editing the InternalSettings.xml file?

Order of elements should not matter as long as they are in the same node.

I think you need to debug one issue at a time. Suggest starting with Task "hanging". Is the process still running?
Henrik
Support
http://www.visualcron.com 
Please like  VisualCron on facebook!
rprastein
2016-11-18T18:26:49Z
Originally Posted by: Support 

Why do you edit the XML files? Are you only editing the InternalSettings.xml file?

Order of elements should not matter as long as they are in the same node.

I think you need to debug one issue at a time. Suggest starting with Task "hanging". Is the process still running?



Hi, Henrik.

I'm editing the XML files because I don't have a tool to diff and merge jobs (or other VC objects). I develop the jobs, variables, connections, notifications, etc. on one server, push them to a second server for integration testing, and then plan to push to a new production server where they will serve my internal customers. With a text file description of the jobs and other objects, I can do a comparison and see what differences there are, determine which ones need to stay different because the server is different, and which ones I need to push from one server to the next. Using the GUI this is very difficult, time consuming, and error prone. I need something text-based and hierarchical.

I don't need to make any changes to the internal settings at this point, and the changes to credentials, connections, etc. will be few and mostly additions. The changes to jobs, however, are frequent, and can be either major or subtle.

To the current problem: I thought that I had a work-around, and everything was working OK at the time I last posted yesterday. Today, however, I see in DEV I have
- two jobs with a task that has status sticking at 99% in the GUI (one for 7.5 hours and one for 21.5 hours execution time)
- one job with a task that shows status as waiting, but result as running, with execution time 0.1 seconds and last run yesterday about 23 hours ago, next scheduled run in in about an hour
- one job where all of the tasks have a status of waiting and result of success, and the job itself has a status of waiting but has a result of running.
in VAL I have
- three jobs with a task that has status sticking at 99% in the GUI (for 1.5 hours, 8 minutes, and 22 minutes execution times)

Clicking Running Jobs shows no jobs currently running on either server.
The tasks that have a status of Running - 99% all have a result of Success and all have a last run that is less than a second later than the last run of the preceeding task. I've seen it turn over, and the previous task runs, then the task in question briefly blinks an execution time of 0 and then goes back to continuing from whatever it was counting from before.

I don't know whether or not it matters, but all of the affected tasks interact with the filesystem - all of the ones that are currently stuck at 99% are logging output from the previous task, and are the second task in a two-task job.

Thanks for your help,

Rebeccah
rprastein
2016-11-18T18:32:51Z
I will send you an export of my DEV settings files.
Scroll to Top