Copy link to clipboard
Copied
Hi, I'm currently researching on why my company's ColdFusion application's neo-cron.xml file occasionally gets wiped out. This issue only happens to certain clients, so we're suspecting that it's an environmental issue and not something (at least not directly) caused by our application. We have recreated this issue occasionally on our internal testing/development environments, and both times it seems that the ColdFusion server is running low on resources (i.e. available memory is running low, processing is nearly reaching 100%, etc.). I have attempted to search elsewhere for possible causes for it, but had no luck. So I'm hoping the Adobe community has some insight into this problem. Thanks!
Copy link to clipboard
Copied
Hello,
I've had the neo-cron file corrupt itself a few times on one particular server.
If you open the file is it just blank?
For us, this was caused by a scheduled task that a customer was entering using CFSCHEDULE. I dont know what they did to break it, but break it they did, and continue to do so... I keep a backup and over-write when it happens.
If you have the same type of issue i would be interested in the fix if you ever get one.
Cheers
Luke
Copy link to clipboard
Copied
Hi Luke,
Can you provide some details on the server specification where the neo-cron file got corrupted (and yep, the file just got wiped out on our servers)? Also, what types of operation does this scheduled task involve? It would be helpful to gather that information so maybe some theory can be made about what type of scheduled task would corrupt the neo-cron file on the server. But the scheduled task of the customer doesn't always corrupt the file, right?
Thanks,
Wai
Copy link to clipboard
Copied
We just got word of another instance of this - with ACF2021 (probably with latest patches, but I can't say for certain). One of our customers got an error while executing a <CFSchedule> tag, indicating that the file extension (for the output file) was invalid. On inspection, neo-cron.xml contained about 30K of nulls (and nothing else)
The only possible 'news' here is that the errors started showing up only minutes after the ColdFusion service was restarted. There's no other evidence that the startup was the key to the problem, but the correlation could be meaningful.
My personal guess is that there's a resource conflict between the logic behind CFSchedule and the cron task - and that this conflict leads to the corruption when the two threads line up in exactly the wrong way.
Copy link to clipboard
Copied
While your current experience may have nothing to do with those of folks in this thread from 11 years ago, I'll assert that a different explanation would be if the previous run of cf had crashed (or been killed while running), while cf was trying to write to that neo-cron.xml file...which it may do for reasons other than creating, editing, or deleting a task. I've observed it editing the file merely when tasks run.
Sadly, only restoring from some backup you tool will recover the file and tasks. Cf only keeps a single generation backup, which it readily overwrites for the same reason of frequent activity attempted to the file. I'd raised a tracker ticket about this some years ago, FWIW.