CF2021 Update 21 - Errors with schedule tasks - ISVALIDPUBLISHPATH is undefined in TASK

[Aaron000001]

Since updating to update 21 from update 20, opening Scheudle Tasks in cfadmin is erroing on some scheduled tasks that were created before update 21. 

The impacted tasks can be edited and resaved which will clear the error.

New scheduled tasks can be created and appear to be working as normal. 

The error we're seeing for impacted tasks: 

Element ISVALIDPUBLISHPATH is undefined in TASK.

The error occurred in scheduletasks.cfm: line 700
Called from scheduletasks.cfm: line 568
Called from scheduletasks.cfm: line 567
Called from scheduletasks.cfm: line 564
Called from scheduletasks.cfm: line 524
Called from scheduletasks.cfm: line 482
Called from scheduletasks.cfm: line 1

[BKBK]

First thing to try before we look further:

1.  Stop ColdFusion.
2.  Empty the directory /bin/felix-cache. That is, delete everything from it.
3.  Restart ColdFusion.
4.  Check whether the issue has been resolved.

These steps resolve the majority of issues resulting from recent ColdFusion updates.

[BKBK]

In addition, the release notes for ColdFusion 2021 Update 21 contain the following warning:

If you have added custom entries to the pathfilter.json file for scheduled task output file allowlisting, you must either:

• Back up the pathfilter.json file before applying the update, or

• Restore it from the update backup after applying the update.

The backup copy of the file can be found at:
          \hf-updates\hf-2021-00021-330446\backup\lib\pathfilter.json

[Charlie Arehart]

Just a point of clarification: you can just tell people to simply delete the felix-cache folder rather than "delete everything from it". Subtle difference, but it's fewer clicks/keystrokes. My main point is to clarify that there's no reason NOT to just delete the folder. Cf will recreate it on restart.

If you see it as "a distinction without a difference", that's fine and no need to reply. As always, I'm just trying to help, not meaning to denigrate. 

[BKBK]

Yes, I see it as "a distinction without a difference". 🙂 It works either way.

[Thomassssss]

I believe I am having the same issue described above. 

I deleted the felix-cache already while fixing and issue discussed and explained by @Charlie Arehart here: 

https://community.adobe.com/t5/coldfusion-discussions/cf2021-cfmail-error-after-update-21/m-p/154096... 

However, this has not fixed the issue seen in my cfadmin Schedule Tasks area.  

From my testing it appears that new cfschedules are still running correctly and as expected, however one or more of the schedules created and existed before the update has somehow corrupted.  

I am able to "fix" the issue by reverting /lib/neo-cron.xml back to a state with no set schedules and set no cluster setup, however this isn't going to be a viable solution for our production servers where hundreds of schedules are set that will need to remain after the update. 

Any further assistance you can provide would be greatly appreciated, as this update will need to be made to production soon.  

[Thomassssss]

To add to my above comment, it appears that I only need to remove the cluster setup details from the /lib/neo-cron.xml file in order to fix the issue in the cfadmin area, I can then recreate the cluster.

Again, not really a viable solution for production, however to me this does appear to be related to some disconnect between the cluster setup and ColdFusion? 

[BKBK]

I am able to "fix" the issue by reverting /lib/neo-cron.xml back to a state with no set schedules and set no cluster setup, however this isn't going to be a viable solution for our production servers where hundreds of schedules are set that will need to remain after the update. 

Any further assistance you can provide would be greatly appreciated, as this update will need to be made to production soon.  

By @Thomassssss

From what you describe, it appears that the ColdFusion update gets rid of the existing /lib/neo-cron.xml file, without saving a back-up. Hence the following suggestion:

1.  Obtain a copy of the file  /lib/neo-cron.xml from your production servers. (Here, and in any other tests you do, keep back-up copies of the neo-cron.xml files, each with appropriate name and version).
2.   If the file contains environment-specific data, such as URLs, replace them with those of the test server.
3.   On the test server, replace /lib/neo-cron.xml with this new file. (Again, remember to keep back-ups) .
4.   Restart ColdFusion.

[Thomassssss]

Thank you for the reply, is the idea of the following steps to test if the schedules on production will have the same corrutpion issues by loading them on to our test system, would this not create issues as the cluster datasource will not match? 

When I changed the /lib/neo-cron.xml is wasn't the changes to the schedule details that fixed anything, it was the removal of the cluser datasource.

Thanks again. 

[BKBK]

Ah, thanks for clarifying. The way I had read it, you were only able to fix the issue by reverting to "no set schedules and set no cluster setup".

[Thomassssss]

Thank you for my reply, this was my first assumption however I tested again and removing the cluster setup from the /lib/neo-cron.xml was the only bit that mattered. 

Replied to myself explaining this.