M Jun
unread,Sep 8, 2026, 7:25:26 AMSep 8Sign in to reply to author
Sign in to forward
You do not have permission to delete messages in this group
Either email addresses are anonymous for this group or you need the view member email addresses permission to view the original message
to vocbench-user
A significant amount of data disappeared from a vocabulary in one of our projects during an overnight daily run. It clearly seems that some run, commit or similar process completed “successfully” rather than being interrupted. All instances and usages of some of our own custom classes had been removed as well as, for example mapping relations such as closeMatch and exactMatch. An interrupted run or commit would probably not result in removals that follow such cleear patterns in the data.
The most puzzling part is that in the project history, the Commit ID for this event had no user associated with it. As far as I have seen, all other commits always show the user who made the commit.
The timing coincides with our own update routines and I have investigated almost everything I have been able to check related to VocBench, GraphDB, communication between VB and the database, the operating system and also the server and network traffic. So far, I have not found any clear cause.
Fortunately, I was able to restore the missing data and also make sure that the relevant changes made after the incident were not lost, using the logs written by VocBench on the server. Our vocabulary maintainers / information specialists found the following - btw - a bit inconvenience: in the UI, the list of changes shown for a Commit ID stops at 100 entries. It would be useful if this could be changed to a paginated or scrollable view so that all changes could be inspected.
My next step is to dig into the logging in different parts of the system and also into the benefits, drawbacks and limitations of doing so. This would hopefully give us more information if something similar happens again (hopefylly not). It would of course be much better to find out what caused this in the first place and avoid a repeat of what was quite a serious situation for us, since it put the maintenance and publication of our most important vocabulary at risk.
The main problem is that we still do not know what actually caused the event or whether the cause was related to VocBench, GraphDB or our own systems and automated runs.
Has anyone here experienced anything similar?
VocBench 15.1.0
GraphDB 10.6.4 / RDF4J 4.3.10 / Connectors 16.2.7
Best regards,
Mika
Finto team / National Library of Finland