Resync gets cancelled on all nodes and sites, and I can restart it with bandwith limits.
Actual behavior
Says it is cancelled successfully, but then get blocked by:
mcli admin replicate resync cancel remote local
Site resync with ID f325e1dc-15ff-4008-ace8-e5528e980941 canceled successfully.
mcli admin replicate --limit-upload 10M --limit-download 10M resync start remote local
mcli: <ERROR> Unable to start replication resync. Invalid site-replication request (site replication resync is already in progress).
Steps to reproduce the behavior
This cluster is in production, so I can't take it down to reproduce it.
But more or less, it all started with:
mcli admin replicate resync start remote local
And since we didn't limit the bandwidth, customers noticed slowdowns and halts, so we had to cancel the resync (the remote contained all files we needed over at local, so we setup site replication and started a resync), so that our customers could continue their work, while we issued a new resync with bandwidth limitations, which then didn't work, because it complained that there was already a resync ongoing.
mc --version
mcli version RELEASE.2024-05-24T09-08-49Z (commit-id=a8fdcbe7cb2f85ce98d60e904717aa00016a7d37)
Runtime: go1.22.3 linux/amd64
Copyright (c) 2015-2024 MinIO, Inc.
License GNU AGPLv3 <https://www.gnu.org/licenses/agpl-3.0.html>
Expected behavior
Resync gets cancelled on all nodes and sites, and I can restart it with bandwith limits.
Actual behavior
Says it is cancelled successfully, but then get blocked by:
Steps to reproduce the behavior
This cluster is in production, so I can't take it down to reproduce it.
But more or less, it all started with:
mcli admin replicate resync start remote local
And since we didn't limit the bandwidth, customers noticed slowdowns and halts, so we had to cancel the resync (the
remote
contained all files we needed over atlocal
, so we setup site replication and started a resync), so that our customers could continue their work, while we issued a new resync with bandwidth limitations, which then didn't work, because it complained that there was already a resync ongoing.mc --version