[
https://issues.apache.org/jira/browse/KAFKA-3919?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15355393#comment-15355393
]
Jun Rao commented on KAFKA-3919:
--------------------------------
[~BigAndy], thanks for reporting and investigating this. It seems that somehow
messages with out of order offsets leaked into the broker log, which prevented
the broker to restart. First, a few things about how replication works.
1. At the high level, we try to make replicas byte-wise identical. Each replica
always checkpoints the last committed offsets to a local file. On restarting,
the replica will truncate its local log to the last committed offset before
fetching from the leader. So, in this case, the leader can be a few messages
ahead of the follower. But those messages won't be committed and will be
truncated if the leader becomes the follower.
2. Compressed messages are always appended to the log atomically in both the
leader and the follower. So, during replication, in theory, it shouldn't be
possible for a follower to request an offset that's in the middle of a
compressed message set.
3. The only case that we allow the replicas to diverge a bit is when unclean
leader election is enabled. Suppose that a non-in-sync replica becomes the
leader and has messages from offset 0 to 100. It then committed new messages
from 101 to 200. Now, another previous in-sync replica may come back with a
last committed offset at 150 (which is larger than the last offset when the
leader is started). This replica will only get messages from 151 to 200 from
the new leader. It thinks it already has messages from 101 to 150, but those
are actually different from the new leader's data. The divergence won't be
forever though.
So, I am not sure if your hypothesis can indeed happen. It would be useful to
dig deeper on this. Do you still have the log segment files? If so, could you
use our DumpLogSegments tool to see the layout of the offsets? Also, did you
have unclean.leader.election disabled? When the new leader becomes the
follower, it will log the offset that it truncates to during startup. Could you
dig that out too?
> Broker faills to start after ungraceful shutdown due to non-monotonically
> incrementing offsets in logs
> ------------------------------------------------------------------------------------------------------
>
> Key: KAFKA-3919
> URL: https://issues.apache.org/jira/browse/KAFKA-3919
> Project: Kafka
> Issue Type: Bug
> Components: core
> Affects Versions: 0.9.0.1
> Reporter: Andy Coates
>
> Hi All,
> I encountered an issue with Kafka following a power outage that saw a
> proportion of our cluster disappear. When the power came back on several
> brokers halted on start up with the error:
> {noformat}
> Fatal error during KafkaServerStartable startup. Prepare to shutdown”
> kafka.common.InvalidOffsetException: Attempt to append an offset
> (1239742691) to position 35728 no larger than the last offset appended
> (1239742822) to
> /data3/kafka/mt_xp_its_music_main_itsevent-20/00000000001239444214.index.
> at
> kafka.log.OffsetIndex$$anonfun$append$1.apply$mcV$sp(OffsetIndex.scala:207)
> at kafka.log.OffsetIndex$$anonfun$append$1.apply(OffsetIndex.scala:197)
> at kafka.log.OffsetIndex$$anonfun$append$1.apply(OffsetIndex.scala:197)
> at kafka.utils.CoreUtils$.inLock(CoreUtils.scala:262)
> at kafka.log.OffsetIndex.append(OffsetIndex.scala:197)
> at kafka.log.LogSegment.recover(LogSegment.scala:188)
> at kafka.log.Log$$anonfun$loadSegments$4.apply(Log.scala:188)
> at kafka.log.Log$$anonfun$loadSegments$4.apply(Log.scala:160)
> at
> scala.collection.TraversableLike$WithFilter$$anonfun$foreach$1.apply(TraversableLike.scala:772)
> at
> scala.collection.IndexedSeqOptimized$class.foreach(IndexedSeqOptimized.scala:33)
> at scala.collection.mutable.ArrayOps$ofRef.foreach(ArrayOps.scala:108)
> at
> scala.collection.TraversableLike$WithFilter.foreach(TraversableLike.scala:771)
> at kafka.log.Log.loadSegments(Log.scala:160)
> at kafka.log.Log.<init>(Log.scala:90)
> at
> kafka.log.LogManager$$anonfun$loadLogs$2$$anonfun$3$$anonfun$apply$10$$anonfun$apply$1.apply$mcV$sp(LogManager.scala:150)
> at kafka.utils.CoreUtils$$anon$1.run(CoreUtils.scala:60)
> at
> java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
> at java.util.concurrent.FutureTask.run(FutureTask.java:266)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> {noformat}
> The only way to recover the brokers was to delete the log files that
> contained non monotonically incrementing offsets.
> I’ve spent some time digging through the logs and I feel I may have worked
> out the sequence of events leading to this issue, (though this is based on
> some assumptions I've made about the way Kafka is working, which may be
> wrong):
> Given:
> * A topic that is produced to using acks = 1
> * A topic that is produced to using gzip compression
> * A topic that has min.isr set to less than the number of replicas, (i.e.
> min.isr=2, #replicas=3)
> * Following ISRs are lagging behind the leader by some small number of
> messages, (which is normal with acks=1)
> * brokers are configured with fairly large zk session timeout e.g. 30s.
> Then:
> When something like a power outage take out all three replicas, its possible
> to get into a state such that the indexes won’t rebuild on a restart and a
> broker fails to start. This can happen when:
> * Enough brokers, but not the pre-outage leader, come on-line for the
> partition to be writeable
> * Producers produce enough records to the partition that the head offset is
> now greater than the pre-outage leader head offset.
> * The pre-outage leader comes back online.
> At this point the logs on the pre-outage leader have diverged from the other
> replicas. It has some messages that are not in the other replicas, and the
> other replicas have some records not in the pre-outage leader's log - at the
> same offsets.
> I’m assuming that because the current leader has at higher offset than the
> pre-outage leader, the pre-outage leader just starts following the leader and
> requesting the records it thinks its missing.
> I’m also assuming that because the producers were using gzip, so each record
> is actual a compressed message set, that iwhen the pre-outage leader requests
> records from the leader, the offset it requests could just happened to be in
> the middle of a compressed batch, but the leader returns the full batch.
> When the pre-outage leader appends this batch to its own log it thinks all is
> OK. But what has happened is that the offsets in the log are no longer
> monotonically incrementing. Instead they actually dip by the number of
> records in the compressed batch that were before the requested offset. If
> and when this broker restarts this dip may be at the 4K boundary the indexer
> checks. If it is, the broker won’t start.
> Several of our brokers were unlucky enough to hit that 4K boundary, causing a
> protracted outage. We’ve written a little utility that shows several more
> brokers have a dip outside of the 4K boundary.
> There are some assumptions in there, which I’ve not got around to confirming
> / denying. (A quick attempt to recreate this failed and I've not found the
> time to invest more).
> Of course I'd really appreciate the community / experts stepping in and
> commenting on whether my assumptions are right or wrong, or if there is
> another explanation to the problem.
> But assuming I’m mostly right, then the fact the broker won’t start is
> obviously a bug, and one I’d like to fix. A Kafka broker should not corrupt
> its own log during normal operation to the point that it can’t restart!
> A secondary issue is if we think the divergent logs are acceptable? This may
> be deemed acceptable given the producers have chosen availability over
> consistency when they produced with acks = 1? Though personally, the system
> having diverging replicas of an immutable commit log just doesn't sit right.
> I see us having a few options here:
> * Have the replicas detect the divergence of their logs e.g. a follower
> compares the checksum of its last record with the same offset on the leader.
> The follower can then workout that its log has diverged from the leader. At
> which point it could either halt, stop replicating that partition or search
> backwards to find the point of divergence, truncate and recover. (possibly
> saving the truncated part somewhere). This would be a protocol change for
> Kafka. This solution trades availability, (you’ve got less ISRs during the
> extended re-sync process), for consistency.
> * Leave the logs as they are and have the indexing of offsets in the log on
> start up handle such a situation gracefully. This leaves logs in a divergent
> state between replicas, (meaning replays would yield different messages if
> the leader was up to down), but gives better availability, (no time spent not
> being an ISR while it repairs any divergence).
> * Support multiple options and allow it be tuned, ideally by topic.
> * Something else...
> I’m happy/keen to contribute here. But I’d like to first discuss which option
> should be investigated.
> Andy
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
