guojialiang92 commented on PR #14417:
URL: https://github.com/apache/lucene/pull/14417#issuecomment-2766116736
Thanks, @vigyasharma
I also looked at Lucene's native segment replication, just sharing my
personal opinion.
> Also, IIUC `IndexWriter#advanceSegmentInfosVersion()` was added to handle
similar scenarios for NRT replication (Lucene's native segment replication
implementation). I'm curious why we didn't run into the need to advance
`SegmentInfos#counter` at that time. Do you remember, @mikemccand (I know it's
been a while! (: )?
In the code comments of Lucene's native segment replication, the risk of
file conflicts is also mentioned, but no additional processing is done. From a
robustness perspective, perhaps control should also be carried out. The
relevant code is as follows：
ReplicaNode#fileIsIdentical (**Segment name was reused! This is rare but
possible and otherwise devastating**)
```
private boolean fileIsIdentical(String fileName, FileMetaData srcMetaData)
throws IOException {
FileMetaData destMetaData = readLocalFileMetaData(fileName);
if (destMetaData == null) {
// Something went wrong in reading the file (it's corrupt, truncated,
does not exist, etc.):
return false;
}
if (Arrays.equals(destMetaData.header(), srcMetaData.header()) == false
|| Arrays.equals(destMetaData.footer(), srcMetaData.footer()) ==
false) {
// Segment name was reused! This is rare but possible and otherwise
devastating:
if (isVerboseFiles()) {
message("file " + fileName + ": will copy [header/footer is
different]");
}
return false;
} else {
return true;
}
}
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
