When an online backup is performed, the first step is to backup the catalog tablespace. To ensure consistency, the backup application takes a share lock on all catalog tables.
There are certain operations, like LOAD, RUNSTATS, DDLs etc, that tend to take an X-lock on a row(s) of the system catalog tables. Naturally, such applications will timeout when the catalog is being backed up or the backup may fail if such operations are in progress ...
This is more a timing issue ... I have worked around this by having a retry logic in runstats and/or backup scripts and scheduling the conflicting operations to happen at different times ...
Does this explain your problem or are you seeing something different ???
Cheers
Sathyaran