It has been observed in FTDDC (Full Time Diagnostic Data Capture) charts that if the history store is large, and then a restart happens, there is a clear gap where the history store file size is reported as 0. Then it jumps back up to its former pre-restart size.
The statistic in question is, from dist/stat_data.py:
CacheStat('cache_hs_ondisk', 'history store table on-disk size', 'no_clear,no_scale,size'),
Currently this is only set in wt_hs_insert_updates:
WT_STAT_CONN_SET(session, cache_hs_ondisk, hs_size);
This ticket should add in setting that statistic when we open the history store file on restart. The file may or may not exist yet, but if it does, then we should get the size and set this statistic value early.
Otherwise the value stays at 0 until more/new entries are added to the history store and that does not reflect the actual size of the file until then.