Scan same Record with same PK from different Node. Duplication Dirty Record

1

Hi, all.

I’m using Aerospike 3.15.3.14

Cluster is composed of 22 nodes.

Every day i need to scan all node separately. and write all Record into separately file.

From one day, i found there were two same key in two files( means two nodes, just think node A and B ). And when I use the aql to query only get one value( always meaning get the result from one fixed node, just think get from Node A ), and when i update the record, only the record in Node A will be updated( can see the different from scan file of next day ), and record with same key in Node B cannot be access, cannot query, cannot update, cannot deleted like a dirty record.

I think this case is happened during data migration, but i am not sure about this.

I want to delete the record in Node B.

However, I cannot access this record in Node B.

Thanks

2

Having difficulty following your description. That said, 3.15.3.14 is very old with many known issues, it is also no longer a supported version. I suggest that an upgrade may fix your issue. If you have Enterprise support, open a case with them to help you along. If you are using community, I don’t think many will be willing to invest time investigating such an old build since it is unlikely to be mutually beneficial.

3

Are you able to share a list of your sets name? I wanted to double check that you are not using any forbidden characters like “:” or “;” in the set names.

asinfo -v sets -l

Most likely you have the same PK in different sets. You can check for illegal characters in set names using this guide. https://discuss.aerospike.com/t/how-to-detect-illegal-characters-in-set-names/6742

or running:

asadm --no-config-file -e "asinfo -v sets -l"|sed -e 's/:objects.*//g'

Another possibility is that you have some records with null sets. (no sets specified when they were written). A scan on the namespace would show two records (1 with no set name and 1 with a set name.)

example with two records:

aql> INSERT INTO test.demo (PK, foo, bar) VALUES ('key1', 123, 'abc')
OK, 1 record affected.

aql> INSERT INTO test (PK, foo, bar) VALUES ('key1', 123, 'abc')
OK, 1 record affected.


aql> select * from test.demo
+-----+-------+
| foo | bar   |
+-----+-------+
| 123 | "abc" |
+-----+-------+
1 row in set (0.076 secs)

OK

aql> select * from test
+-----+-------+
| foo | bar   |
+-----+-------+
| 123 | "abc" |
| 123 | "abc" |
+-----+-------+
2 rows in set (0.074 secs)

OK