[
https://issues.apache.org/jira/browse/CASSANDRA-13700?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16094911#comment-16094911
]
Joel Knighton commented on CASSANDRA-13700:
-------------------------------------------
Thanks, Jason! In this case, I agree the first option is safer for this issue.
Something like the second likely makes sense eventually, at least as part of a
larger audit of correctness issues in gossip. I believe your volatile
suggestion is correct.
I don't have a lot of helpful information to reproduce this; it reproduces in
larger clusters, particularly with higher latency levels. We can see the
effects locally with a few well-timed sleeps in MessagingService, but that
isn't terribly representative.
Branches pushed here:
||branch||
|[13700-2.1|https://github.com/jkni/cassandra/tree/13700-2.1]||
|[13700-2.2|https://github.com/jkni/cassandra/tree/13700-2.2]||
|[13700-3.0|https://github.com/jkni/cassandra/tree/13700-3.0]||
|[13700-3.11|https://github.com/jkni/cassandra/tree/13700-3.11]||
|[13700-trunk|https://github.com/jkni/cassandra/tree/13700-trunk]||
There's a somewhat conceptually similar issue when we bump the gossip
generation in the middle of constructing a reply - I believe that's the cause
in [CASSANDRA-11825], which presents similar problems. I'm choosing to address
them separately because they're indeed distinct problems and 11825 requires an
additional trigger (enabling and disabling gossip during runtime).
> Heartbeats can cause gossip information to go permanently missing on certain
> nodes
> ----------------------------------------------------------------------------------
>
> Key: CASSANDRA-13700
> URL: https://issues.apache.org/jira/browse/CASSANDRA-13700
> Project: Cassandra
> Issue Type: Bug
> Components: Distributed Metadata
> Reporter: Joel Knighton
> Assignee: Joel Knighton
> Priority: Critical
>
> In {{Gossiper.getStateForVersionBiggerThan}}, we add the {{HeartBeatState}}
> from the corresponding {{EndpointState}} to the {{EndpointState}} to send.
> When we're getting state for ourselves, this means that we add a reference to
> the local {{HeartBeatState}}. Then, once we've built a message (in either the
> Syn or Ack handler), we send it through the {{MessagingService}}. In the case
> that the {{MessagingService}} is sufficiently slow, the {{GossipTask}} may
> run before serialization of the Syn or Ack. This means that when the
> {{GossipTask}} acquires the gossip {{taskLock}}, it may increment the
> {{HeartBeatState}} version of the local node as stored in the endpoint state
> map. Then, when we finally serialize the Syn or Ack, we'll follow the
> reference to the {{HeartBeatState}} and serialize it with a higher version
> than we saw when constructing the Ack or Ack2.
> Consider the case where we see {{HeartBeatState}} with version 4 when
> constructing an Ack and send it through the {{MessagingService}}. Then, we
> add some piece of state with version 5 to our local {{EndpointState}}. If
> {{GossipTask}} runs and increases the {{HeartBeatState}} version to 6 before
> the {{MessageOut}} containing the Ack is serialized, the node receiving the
> Ack will believe it is current to version 6, despite the fact that it has
> never received a message containing the {{ApplicationState}} tagged with
> version 5.
> I've reproduced in this in several versions; so far, I believe this is
> possible in all versions.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
