Github user tdas commented on a diff in the pull request:
https://github.com/apache/spark/pull/15820#discussion_r88137338
--- Diff:
external/kafka-0-10-sql/src/main/scala/org/apache/spark/sql/kafka010/CachedKafkaConsumer.scala
---
@@ -83,6 +86,139 @@ private[kafka010] case class CachedKafkaConsumer
private(
record
}
+ /**
+ * Get the record at the `offset`. If it doesn't exist, try to get the
earliest record in
+ * `[offset, untilOffset)`.
+ */
+ def getAndIgnoreLostData(
+ offset: Long,
+ untilOffset: Long,
+ pollTimeoutMs: Long): ConsumerRecord[Array[Byte], Array[Byte]] = {
+ // scalastyle:off
+ // When `failOnDataLoss` is `false`, we need to handle the following
cases (note: untilOffset and latestOffset are exclusive):
+ // 1. Some data are aged out, and `offset < beginningOffset <=
untilOffset - 1 <= latestOffset - 1`
+ // Seek to the beginningOffset and fetch the data.
+ // 2. Some data are aged out, and `offset <= untilOffset - 1 <
beginningOffset`.
+ // There is nothing to fetch, return null.
+ // 3. The topic is deleted.
+ // There is nothing to fetch, return null.
+ // 4. The topic is deleted and recreated, and `beginningOffset <=
offset <= untilOffset - 1 <= latestOffset - 1`.
+ // We cannot detect this case. We can still fetch data like
nothing happens.
+ // 5. The topic is deleted and recreated, and `beginningOffset <=
offset < latestOffset - 1 < untilOffset - 1`.
+ // Same as 4.
+ // 6. The topic is deleted and recreated, and `beginningOffset <=
latestOffset - 1 < offset <= untilOffset - 1`.
+ // There is nothing to fetch, return null.
+ // 7. The topic is deleted and recreated, and `offset <
beginningOffset <= untilOffset - 1`.
+ // Same as 1.
+ // 8. The topic is deleted and recreated, and `offset <= untilOffset -
1 < beginningOffset`.
+ // There is nothing to fetch, return null.
+ // scalastyle:on
+ require(offset < untilOffset, s"offset: $offset, untilOffset:
$untilOffset")
+ logDebug(s"Get $groupId $topicPartition nextOffset
$nextOffsetInFetchedData requested $offset")
+ try {
+ if (offset != nextOffsetInFetchedData) {
+ logInfo(s"Initial fetch for $topicPartition $offset")
+ seek(offset)
+ poll(pollTimeoutMs)
+ } else if (!fetchedData.hasNext()) {
+ // The last pre-fetched data has been drained.
+ poll(pollTimeoutMs)
+ }
+ getRecordFromFetchedData(offset, untilOffset)
+ } catch {
+ case e: OffsetOutOfRangeException =>
+ logWarning(s"Cannot fetch offset $offset. Some data may be lost.
Recovering from " +
+ "the beginning offset", e)
+ advanceToBeginningOffsetAndFetch(offset, untilOffset,
pollTimeoutMs)
+ }
+ }
+
+ /**
+ * Try to advance to the beginning offset and fetch again.
`beginningOffset` should be in
+ * `[offset, untilOffset]`. If not, it will try to fetch `offset` again
if it's in
+ * `[beginningOffset, latestOffset)`. Otherwise, it will return null and
reset the pre-fetched
+ * data.
+ */
+ private def advanceToBeginningOffsetAndFetch(
+ offset: Long,
+ untilOffset: Long,
+ pollTimeoutMs: Long): ConsumerRecord[Array[Byte], Array[Byte]] = {
+ val beginningOffset = getBeginningOffset()
+ if (beginningOffset <= offset) {
+ val latestOffset = getLatestOffset()
+ if (latestOffset <= offset) {
+ // beginningOffset <= latestOffset - 1 < offset <= untilOffset - 1
+ logWarning(s"Offset ${offset} is later than the latest offset
$latestOffset. " +
+ s"Skipped [$offset, $untilOffset)")
+ reset()
+ null
+ } else {
+ // beginningOffset <= offset <= min(latestOffset - 1, untilOffset
- 1)
+ //
+ // This will happen when a topic is deleted and recreated, and new
data are pushed very fast
+ // , then we will see `offset` disappears first then appears
again. Although the parameters
+ // are same, the state in Kafka cluster is changed, so it's not an
endless loop.
+ //
+ // In addition, the stack here won't be deep unless the user keeps
deleting and creating the
+ // topic very fast.
+ //
+ // Therefore, this recursive call is safe.
+ getAndIgnoreLostData(offset, untilOffset, pollTimeoutMs)
+ }
+ } else {
+ if (beginningOffset >= untilOffset) {
+ // offset <= untilOffset - 1 < beginningOffset
+ logWarning(s"Buffer miss for $groupId $topicPartition [$offset,
$untilOffset)")
+ reset()
+ null
+ } else {
+ // offset < beginningOffset <= untilOffset - 1
+ logWarning(s"Buffer miss for $groupId $topicPartition [$offset,
$beginningOffset)")
+ getAndIgnoreLostData(beginningOffset, untilOffset, pollTimeoutMs)
+ }
+ }
+ }
+
+ /**
+ * Get the earliest record in [offset, untilOffset) from the fetched
data. If there is no such
+ * record, returns null. Must be called after `poll`.
--- End diff --
This is not true. This sounds like poll() has to be called immediately
before this method is called. This is not the case in the above usage where
poll is called only if the conditions match.
Rather say "This must be called after some data has already been fetched
using `poll`."
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
