Object count delta between src and dst bucket within a Snapmirror-S3 relationship

Last updated
Save as PDF
Share
1. Share
2. Tweet
3. Share

Views:: 191

Visibility:: Public

Votes:: 0

Category:: ontap-9

Specialty:: dp

Last Updated:

Applies to

Snapmirror-S3 (FabricLink)
ONTAP_S3 object store bucket

Issue

Within a Snapmirror-S3 FabricLink relationship with RPO=0, the destination object-store bucket object-count keeps on raising, possibly way above the object-count of the source bucket.
Aggregate hosting the affected bucket keeps on filling and space is getting tight or aggregate runs out of space.
Alternatively, in a lighter form, the object-count between source- and destination-bucket within the Snapmirror-S3 relationship, shows only different numbers with no outstanding work:

::> vserver object-store-server bucket show -fields object-count vserver bucket object-count ------- ---------- ------------ svm_dst bucket-dst 124080 svm_src bucket-src 124078 2 entries were displayed.

::> snapmirror show Progress Source Destination Mirror Relationship Total Last Path Type Path State Status Progress Healthy Updated ----------- ---- ------------ ------- -------------- --------- ------- -------- svm_src:/bucket/bucket-src XDP svm_dst:/bucket/bucket-dst Snapmirrored Idle - true -

::*> snapmirror show -instance Source Path: svm_src:/bucket/bucket-src Destination Path: svm_dst:/bucket/bucket-dst Relationship Type: XDP SnapMirror Policy Type: continuous SnapMirror Policy: Continuous Throttle (KB/sec): unlimited Mirror State: Snapmirrored Relationship Status: Idle Relationship ID: 16d619e6-9044-11ed-823e-005056bbfb01 FabricLink Source Role: bucket-src FabricLink Source Bucket: active-mirror FabricLink Peer Role: bucket-dst FabricLink Peer Bucket: active-mirror FabricLink Topology: 145c48be-9044-11ed-8707-005056bbcabb FabricLink Pull Byte Count: 0 FabricLink Push Byte Count: 0 FabricLink Pending Work Count: 0 FabricLink Status: ok

After manual deletion of all objects within the source-bucket, the object-count either on source-bucket, or on destination-bucket, or on both, does not drop down to zero and some or even the majority of occupied space is not released.
Deeper investigation shows, partial multipart-objects are left within the destination bucket when checking from an external S3 client like aws cli:

$ aws --profile bucket-dst --endpoint-url https://ontap_s3.local s3api list-multipart-uploads --bucket bucket-dst --output text UPLOADS 2023-01-09T09:57:09.000Z f001 MjE1NTI5NTY5OF8xMDU2XzIzMF80NjgzNzU4Mg UPLOADS 2023-01-09T10:03:46.000Z f002 MjE1NTI5NTY5N18xMDU1XzE3MV80NjQzNTk2MQ

Note: If the destination-bucket had ever been promoted (Snapmirror-S3 was reversed), partial multipart objects can remain in both, source- and destination-bucket.

At the time when the Snapmirror-S3 relationship was initialized, air.repaired.vplus:notice events are reported within ONTAP event log show (EMS) output:

Thu Sep 15 09:07:09 +0200 [zbghns4-03: wafl_exempt07: air.repaired.vplus:notice]: AIR vplus rebuild of 1204~0.147527 completed for subtype 6