CODEWITHSUNDEEP
aerospike
Paul Leclercq
Paul Leclercq

Reputation: 1018

Asrestore : secondary indexes get deleted when already existing

When restoring a backup of a cluster (3.8.2.3) into the same cluster, secondary indexes get deleted :

show indexes Namespace gives 36 rows

cat Backup-Namespace.abs.gz | gunzip | asrestore --input-file - --no-records

show indexes Namespace gives 0 rows

And if I restore once again with the same command show indexes Namespace gives 36 rows

Same result for the option --replace, --unique

It works for --no-generation option, but as I would like the the lastest data, I don't want to use this option.

Note that the backup was created with :

asbackup --no-udfs --priority 1 --namespace Namespace --output-file -

To solve this problem I create indexes by hands where there are new ones, and I use --no-indexes for asrestore to update my data.

Am I doing something wrong ?

Upvotes: 2

Views: 177

Answers (1)

Meher
Meher

Reputation: 2939

You are doing things right. This is an issue with the current asrestore tool. The asrestore tool first drops the indices from the backup using the C client's asynchronous API and doesn't wait for completion before attempting to re-create those same indices. This will cause the re-creation to often fail (since those indices still exist and are still being deleted).

Using the --wait option on the restore should help. And keep an eye for Jira TOOLS-608 in a soon to come Tools package release for some enhancements regarding this odd behavior.

Upvotes: 3

Related Questions