I'm +1 for the change. I think the main advantage is that it simplifies the
batching. More like the LeaderAndIsr/UpdateMetadata requests, the request
grouping becomes less significant. I'm not sure how much of the benefit can
be realized given the need to keep compatibility, but it still seems like
I shouldn't have said "always", sorry.
When a replica is deleted, either due to a topic deletion or a reassignment,
the controller transitions the replica to the `OfflineReplica` state and
then to
the `ReplicaDeletionStarted` state. The first transition issues a
StopReplicaRequest
with DeleteParti
Is it really true that the controller always sends two requests? Aren't the
operations different (stop replica with delete versus stop replica without)?
On Wed, Mar 25, 2020, 9:59 AM David Jacot wrote:
> Hi all,
>
> I'd like to inform you that I have slightly changed the schema which was
> propo
Hi all,
I'd like to inform you that I have slightly changed the schema which was
proposed
in the KIP. During the implementation, I have realized that the proposed
schema
did not work. The new one reorganises how topics/partitions are stored.
I'd like to amend the current KIP with the following:
Hi all,
The vote has passed with +3 binding votes (Jason Gustafson, Gwen Shapira,
Jun Rao).
Thanks to everyone!
Best,
David
On Wed, Mar 4, 2020 at 9:02 AM David Jacot wrote:
> Hi Jun,
>
> You're right. I have noticed it while implementing it. I plan to use a
> default
> value as a sentinel in
Hi Jun,
You're right. I have noticed it while implementing it. I plan to use a
default
value as a sentinel in the protocol (e.g. -2) to cover this case.
David
On Wed, Mar 4, 2020 at 3:18 AM Jun Rao wrote:
> Hi, David,
>
> Thanks for the KIP. +1 from me too. Just one comment below.
>
> 1. Regar
Hi, David,
Thanks for the KIP. +1 from me too. Just one comment below.
1. Regarding the sentinel leader epoch to indicate topic deletion, it seems
that we need to use a different sentinel value to indicate that the leader
epoch is not present when the controller is still on the old version during
+1
On Mon, Feb 24, 2020, 2:16 AM David Jacot wrote:
> Hi all,
>
> I would like to start a vote on KIP-570: Add leader epoch in
> StopReplicaRequest
>
> The KIP is here:
>
> https://cwiki.apache.org/confluence/display/KAFKA/KIP-570%3A+Add+leader+epoch+in+StopReplicaRequest
>
> Thanks,
> David
>
+1
On Mon, Feb 24, 2020 at 2:16 AM David Jacot wrote:
> Hi all,
>
> I would like to start a vote on KIP-570: Add leader epoch in
> StopReplicaRequest
>
> The KIP is here:
>
> https://cwiki.apache.org/confluence/display/KAFKA/KIP-570%3A+Add+leader+epoch+in+StopReplicaRequest
>
> Thanks,
> David
>
Hi all,
I would like to start a vote on KIP-570: Add leader epoch in
StopReplicaRequest
The KIP is here:
https://cwiki.apache.org/confluence/display/KAFKA/KIP-570%3A+Add+leader+epoch+in+StopReplicaRequest
Thanks,
David
10 matches
Mail list logo
