Reputation: 35129
I have two nodes 8xl cluster. And today I've decided to take a look at some metrics that Amazon provides, what I've noticed is that some disks are empty.
From Amazon docs:
capacity integer Total capacity of the partition in 1 MB disk blocks.
SQL:
select owner, used, tossed, capacity, trim(mount) as mount
from stv_partitions
where capacity < 1;
owner | used | tossed | capacity | mount
-------+------+--------+----------+-----------
0 | 0 | 1 | 0 | /dev/xvdo
1 | 0 | 1 | 0 | /dev/xvdo
(2 rows)
Can someone explain to me why am I seeing this? Is that an expected behaviour?
owner | host | diskno | part_begin | part_end | used | tossed | capacity | reads | writes | seek_forward | seek_back | is_san | failed | mbps | mount
-------+------+--------+---------------+---------------+------+--------+----------+-------+--------+--------------+-----------+--------+--------+------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
1 | 1 | 13 | 0 | 1000126283776 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | /dev/xvdo
0 | 1 | 13 | 1000126283776 | 2000252567552 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | /dev/xvdo
Upvotes: 0
Views: 612
Reputation: 2277
It is due to the fact that the device has failed (=1) and hence the disk capacity is set to 0.
Upvotes: 1