[
https://issues.apache.org/jira/browse/RATIS-498?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16807366#comment-16807366
]
Tsz Wo Nicholas Sze commented on RATIS-498:
-------------------------------------------
How about we modify the proto as below (please feel free to modify it)?
{code:java}
message InstallSnapshotRequestProto {
message SnapshotChunkProto {
string requestId = 1; // an identifier for chunked-requests.
uint32 requestIndex = 2; // the index for this request chunk. Starts from 0.
RaftConfigurationProto raftConfiguration = 3;
uint64 leaderTerm = 4;
TermIndexProto termIndex = 5;
repeated FileChunkProto fileChunks = 6;
uint64 totalSize = 7;
bool done = 8; // whether this is the final chunk for the same req.
}
message NotificationProto {
TermIndexProto firstAvailableTermIndex = 1; // first available log index to
notify Follower to install snapshot
}
RaftRpcRequestProto serverRequest = 1;
oneof InstallSnapshotRequestBody {
SnapshotChunkProto snapshotChunk = 2;
NotificationProto notification = 3;
}
}
{code}
When a follower receives an InstallSnapshotRequestProto, if snapshot
installation is disabled, it checks whether the proto has a notification. If
snapshot installation is enabled, it checks whether the proto has a
snapshotChunk.
{quote}And, if leader and follower have different settings for
installSnapshotEnabled, I am thinking we could return a new
InstallSnapshotResult result code (CONF_MISMATCH) to the leader. But there
would not be a way to fix this error, right? Should the follower be shutdown in
this case?
{quote}
Returning CONF_MISMATCH sounds good. I think the follower should not be
shutdown since it could be the case that the leader is misconfigured. Let's
just print a big warning for now.
BTW, since this sounds quite involved, how about we work on it in a separated
JIRA?
> Notify Follower to Install Snapshot through state machine
> ---------------------------------------------------------
>
> Key: RATIS-498
> URL: https://issues.apache.org/jira/browse/RATIS-498
> Project: Ratis
> Issue Type: New Feature
> Components: server
> Reporter: Hanisha Koneru
> Assignee: Hanisha Koneru
> Priority: Major
> Attachments: RATIS-498.000.patch, RATIS-498.001.patch,
> RATIS-498.002.patch, RATIS-498.003.patch, RATIS-498.004.patch,
> RATIS-498.005.patch, RATIS-498.006.patch, RATIS-498.007.patch,
> RATIS-498.008.patch
>
>
> When a lagging Follower wants to catch up with the Leader, and the Leader
> only has logs with start index greater than the Followers's last log index,
> then the leader sends an InstallSnapshotRequest to the the Follower.
> The aim of this Jira is to allow State Machine to decouple snapshot
> installation from the Ratis server. When Leader does not have the logs to get
> the Follower up to speed, it should notify the Follower to install a snapshot
> (if install snapshot through Log Appender is disabled). The Follower in turn
> notifies its state machine that a snapshot is required to catch up with the
> leader.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
