I have a really nasty issue that has occurred twice now. If I restart the gateway server multiple UDT tags have 90% of their member deleted!
I have only rebooted these machines twice and its occurred both times.
When this occurred previously I though it was this issue here and at the time I was running 8.1.7 and the post mentioned there was a fix in 8.1.8 so I upgraded to 8.1.11 (which is what I am on today). Both then and now the “workaround” mentioned in that post does not work (restart tag). I am starting to suspect this is a whole different issue to that post.
Last time to “recover” I had to delete the affected UDT tags and re-create them. This seems to only affect my UDT “SP_IRBL_AC_AND_TEMP”.
I am running:
Maker edition 8.1.11
MeriaDB 1:10.3.31
Both are running on separate Debian 4.19.208-1 instances
I have exported two instances of the UDT. “KatesAC” was unaffected but “OfficeAC” that uses the same typeID has basically nothing in it.
There is also an export of the UDT. The UDT is basically an interface to a MQTT controller for my aircons, there is quite a bit going on there.
I know I went a bit crazy with this UDT but it all works if I dont reboot, I just have NFI why ignition would nuke it on a reboot???
I am going to get it back to a working state, export the tags, update it to 8.1.14 (on the off chance there is an undocumented fix), probably export the tags again and then reboot to see if it happens again…
If anyone knows something about this in the mean time please feel free to share. thanks
UPDATE 1:
I remember that I have daily global backs setup. I tried restoring from backup and it turns out all my backups are affected or the act or “restarting” at the end of the restore causes the issue to occur again.
This issue can happen for a couple of reasons and working with support would be the best way to attempt to troubleshoot and understand the problem. The two most common reasons this occurs is:
We rank all members of a UDT Definition so that they are loaded in the correct manner. If this ranking is incorrect, then members don’t get loaded properly.
If something in the definition of the member tag gets corrupted (like an invalid value written via a script) then on restart the tag is flagged as invalid and won’t start.
Both of these issues typically only show themselves after a tag has been restarted and should have messages in the logs on tag provider startup to help identify where there is a problem. There have been a few Tag Ranking bugs fixed (most recent in 8.1.11).
I would recommend calling into support to work through the issue you are seeing as fixing UDTs via the DB isn’t as simple as it may seem.
That is on me for not realizing the forum I was in. Any chance you can direct message me a way to get access to a copy of backup of your Gateway (via Dropbox or something similar)? Unfortunately the import files aren’t helping us track down the issue.
Upgrading to the latest version (at the time was 8.1.14) fixed my issue.
After the upgrade all my tags where back without any missing members, I did not need to restore any backups it was just fixed.
I want to give the solution to Garth as he did follow up with me via DM and was very helpfull, but none of his public posts suggest an upgrade. In anycase I was going to upgrade to the latest anyway.
I’m having the same issue after a server restart, a few of the subfolders of my UDT now contain no tags. I’ve tried restarting tags and upgrading to the latest version (8.1.18). Nothing has fixed my issue. Checked the logs and it looks like the tag provider is starting normally…
