On a mixed-version cluster, say firefly and dumpling, the first round of
data health checks could end up with crazy values being reported for
data usage/availability for dumpling monitors.
This would be caused by dumpling not supporting reporting of store
stats, and by not assuming values as zero on decoding we would end up
decoding trash.
Signed-off-by: Joao Eduardo Luis <joao@redhat.com>
uint64_t bytes_misc;
utime_t last_update;
+ LevelDBStoreStats() :
+ bytes_total(0),
+ bytes_sst(0),
+ bytes_log(0),
+ bytes_misc(0)
+ {}
+
void dump(Formatter *f) const {
assert(f != NULL);
f->dump_int("bytes_total", bytes_total);