mem_used in increasing and dropping in basic setup with 5 buckets

2 nodes, 5 buckets, 200K x 1KB docs per bucket, 2K updates per bucket.

You can see that mem_used for bucket-1 increased from ~600M to ~1250MB after 5 hours.

It doesn't look like a fragmentation issue, at least allocator stats don't indicate that:

MALLOC: 1575414200 ( 1502.4 MiB) Bytes in use by application
MALLOC: + 24248320 ( 23.1 MiB) Bytes in page heap freelist
MALLOC: + 77763952 ( 74.2 MiB) Bytes in central cache freelist
MALLOC: + 3931648 ( 3.7 MiB) Bytes in transfer cache freelist
MALLOC: + 27337432 ( 26.1 MiB) Bytes in thread cache freelists
MALLOC: + 7663776 ( 7.3 MiB) Bytes in malloc metadata
MALLOC: ------------
MALLOC: = 1716359328 ( 1636.8 MiB) Actual memory used (physical + swap)
MALLOC: + 1581056 ( 1.5 MiB) Bytes released to OS (aka unmapped)
MALLOC: ------------
MALLOC: = 1717940384 ( 1638.4 MiB) Virtual address space used
MALLOC:
MALLOC: 94773 Spans in use
MALLOC: 36 Thread heaps in use
MALLOC: 8192 Tcmalloc page size

Please notice that actual RAM usage (RSS) is pretty stable.

Another issue is dropping mem_used for bucket-2 and bucket-4 and these errors:

Mon Jul 7 10:24:59.559952 PDT 3: (bucket-2) Total memory in memoryDeallocated() >= GIGANTOR !!! Disable the memory tracker...
Mon Jul 7 10:54:58.109779 PDT 3: (bucket-4) Total memory in memoryDeallocated() >= GIGANTOR !!! Disable the memory tracker...
Mon Jul 7 10:54:58.109779 PDT 3: (bucket-4) Total memory in memoryDeallocated() >= GIGANTOR !!! Disable the memory tracker...