Validate source of an index in LuceneChangesSnapshot #32288
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Today we make _source optional in the LuceneChangesSnapshot then validate it in ShardChangesAction. This approach, however, prevents deletes and no-ops from replicating in CCR because accessing _source is forbidden in Deletes and Noops. Moreover, _source should be required for an index. This change checks _source in LuceneChangesSnapshot and fails a read request if _source is not available.
dnhatn
added
>bug
:Distributed Indexing/CCR
|
Pinging @elastic/es-distributed |
dnhatn
changed the title
dnhatn
changed the title
bleskes
approved these changes
Jul 24, 2018
| @@ -258,10 +258,19 @@ private TopDocs searchOperations(ScoreDoc after) throws IOException { | |||
| assert assertDocSoftDeleted(leaf.reader(), segmentDocID) : "Delete op but soft_deletes field is not set [" + op + "]"; | |||
| } else { | |||
| final BytesReference source = fields.source(); | |||
| if (source == null) { | |||
| if (requiredFullRange) { | |||
There was a problem hiding this comment.
why is this tied to requiredFullRange so deep?
PS - I thought we were heading towards always requiring full ranges? this make this problem moot.
There was a problem hiding this comment.
CCR always requires full-range but recovery does not require full-range.
why is this tied to requiredFullRange so deep?
Good question. I will think about it.
|
Recovery also requires a full range, we don’t enforce it on this level
because of simplicity. Check that we wait for the local checkpoint to go
above the max seq no we sample and that we validate it when shipping.
…On Tue, 24 Jul 2018 at 7:26 PM, Nhat Nguyen ***@***.***> wrote:
***@***.**** commented on this pull request.
------------------------------
In
server/src/main/java/org/elasticsearch/index/engine/LuceneChangesSnapshot.java
<#32288 (comment)>
:
> @@ -258,10 +258,19 @@ private TopDocs searchOperations(ScoreDoc after) throws IOException {
assert assertDocSoftDeleted(leaf.reader(), segmentDocID) : "Delete op but soft_deletes field is not set [" + op + "]";
} else {
final BytesReference source = fields.source();
+ if (source == null) {
+ if (requiredFullRange) {
CCR always requires full-range but recovery does not require full-range.
why is this tied to requiredFullRange so deep?
Good question. I will think about it.
—
You are receiving this because your review was requested.
Reply to this email directly, view it on GitHub
<#32288 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AA9bJ6QQW-xmlOgPCG7gFmo0h3aYGaCxks5uJ0qngaJpZM4VbT50>
.
|
|
@bleskes I see what you meant. Once this TODO https://github.com/elastic/elasticsearch/blob/ccr/server/src/main/java/org/elasticsearch/indices/recovery/RecoverySourceHandler.java#L167 gets in, I will remove this leniency. |
|
@elasticmachine test this please |
|
@elasticmachine retest this please |
jasontedor
added a commit
to jasontedor/elasticsearch
that referenced
this pull request
Jul 31, 2018
* elastic/ccr: (57 commits) ShardFollowNodeTask should fetch operation once (elastic#32455) Do not expose hard-deleted docs in Lucene history (elastic#32333) Tests: Fix convert error tests to use fixed value (elastic#32415) IndicesClusterStateService should replace an init. replica with an init. primary with the same aId (elastic#32374) REST high-level client: parse back _ignored meta field (elastic#32362) [CI] Mute DocumentSubsetReaderTests testSearch Reject follow request if following setting not enabled on follower (elastic#32448) TEST: testDocStats should always use forceMerge (elastic#32450) TEST: avoid merge in testSegmentMemoryTrackedInBreaker TEST: Avoid deletion in FlushIT AwaitsFix IndexShardTests#testDocStats Painless: Add method type to method. (elastic#32441) Remove reference to non-existent store type (elastic#32418) [TEST] Mute failing FlushIT test Fix ordering of bootstrap checks in docs (elastic#32417) [TEST] Mute failing InternalEngineTests#testSeqNoAndCheckpoints Validate source of an index in LuceneChangesSnapshot (elastic#32288) [TEST] Mute failing testConvertLongHexError bump lucene version after backport Upgrade to Lucene-7.5.0-snapshot-608f0277b0 (elastic#32390) ...
dnhatn
added a commit
that referenced
this pull request
Aug 2, 2018
Today it's possible to encounter an Index operation in Lucene whose _source is disabled, and _recovery_source was pruned by the MergePolicy. If it's the case, we create a Translog#Index without source and let the caller validate it later. However, this approach is challenging for the caller. Deletes and No-Ops don't allow invoking "source()" method. The caller has to make sure to call "source()" only on index operations. The current implementation in CCR does not follow this and fail to replica deletes or no-ops. Moreover, it's easier to reason if a Translog#Index always has the source.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Today it's possible to encounter an Index operation in Lucene whose _source is disabled, and _recovery_source was pruned by the MergePolicy. If it's the case, we create a Translog#Index without source and let the caller validate it later. However, this approach is challenging for the caller.
Deletes and No-Ops don't allow invoking "source()" method. The caller has to make sure to call "source()" only on index operations. The current implementation in CCR does not follow this and fail to replica deletes or no-ops.
It's easier to reason if a Translog#Index always has the source.