Hi, I’m having 1 under replicated range in the cluster (Cluster Size 7 with version:19.2.2). Recently a node was decommissioned . Output from sql:SHOW EXPERIMENTAL_RANGES for affected table shows that particular range is still mapping one of its replica to the decommissioned node. Restarting the active nodes having this range and adding a new node didn’t help. Is there a way to move the range to another node (with EXPERIMENTAL_RELOCATE?) or how to resolve this issue. Thanks.
If your cluster has an under-replicated range, you can fix it by putting the range through the manual replication queue:
(1) In the Admin UI, click the Gear icon on the left to go to the
Advanced Debugging page
(2) In the
Tracing and Profiling Endpoints (local node only) section, click
Run a range through an internal queue
RangeID=<your range id>
(5) Attach the
Enqueue Range Output section to this post. This section will either show a successful replication or give an explanation as to why the range cannot replicate.
If the manual replication is not successful, please also run and attach the
Range Status report for that particular range. This report can be found on the
Advanced Debugging page in the Admin UI in the
Even More Advanced Debugging section. Of particular interest on the
Range Status report is the value for
MVCC System Bytes/Count.
Hope this info helps!
Technical Support Engineer
Thanks Florence. I ran the a range through an internal queue for problem range 17006. Earlier, the range status was showing invalid lease for old decommissioned node. It’s now transferred to a new node and status updated to under replicated/slow for leader
@florence-crl Thanks for the response. Did the suggested changes but range is still under replicated/slow for lease holder. Range status for this problem range has been attached in previous response.