94 record table got 5.1GB size

I have table with UUID, BLOB, BIGINT
BLOB in avg is 600 bytes and max size 32-64k

CREATE TABLE mydata (
id uuid PRIMARY KEY,
data bytea,
shard_id bigint
);

– Indices -------------------------------------------------------

CREATE INDEX shard_id_idx ON mydata(shard_id);
CREATE UNIQUE INDEX primary ON mydata(id);

It is development deployment but cockroach started crashing and I found our that one of table got a huge size 5.1GB

In logs I see a periodic compaction is done by server. But why one table grow that much. It could be 1MB but not 5GB

cockroach is deployed in in-house kubernets with persistent storage on host machine.

Any advice are welcome.
It was 2.0.9 but today after I found issue I updated to february 2019 latest 2.1.15

Thanks

Can you provide us logs of the crash? Where did you see that the table was 5.1G in size? Is that size surprising? Are you doing lots of INSERTs? What happened after you upgraded? Did it restart fine? Has it crashed again?

I have not see actual error in logs but kubernetes often shows cockroach restarted or just not working after not able to start

NAME READY STATUS RESTARTS AGE APP
cockroachdb-0 1/1 Running 207 1d cockroachdb
cockroachdb-1 1/1 Running 155 1d cockroachdb
cockroachdb-2 1/1 Running 227 1d cockroachdb

You can see restarts are 207, 155, 227