[
https://issues.apache.org/jira/browse/BOOKKEEPER-201?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13267416#comment-13267416
]
Flavio Junqueira commented on BOOKKEEPER-201:
---------------------------------------------
bq. What if Bookie restarts? It may consider back again right.
The idea in this jira targets a single disk. The general idea is that a bookie
might be mostly fine, except for a disk that is underperforming. Perhaps
instead of giving the operator only the option of removing the bookie
completely, we could give this option that enables the bookie to eventually
stop writing to that disk. We probably also want to give the operator the
chance to add disks, just so that we can bring the bookie back to its original
state in the case we replace the disk.
bq. operator can not remove that disk from config as there might be some data
written already.
One idea is to use the bookie recovery tool to reconstruct the ledger fragments
on the faulty disk in another bookie.
> Interface to remove slow disks
> ------------------------------
>
> Key: BOOKKEEPER-201
> URL: https://issues.apache.org/jira/browse/BOOKKEEPER-201
> Project: Bookkeeper
> Issue Type: Improvement
> Reporter: Flavio Junqueira
>
> As per a conversation with Ivan and Ben, we propose an interface that enables
> an operator to make a bookie disk inactive. The general idea is that if a
> bookie has multiple ledger disks and one is continuously underperforming,
> then we would like to tell the bookie to stop creating ledgers on that disk.
> Eventually the bookie stops using that disk.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
