lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Shai Erera (JIRA)" <>
Subject [jira] Updated: (LUCENE-2537) FSDirectory.copy() impl is unsafe
Date Fri, 23 Jul 2010 12:45:50 GMT


Shai Erera updated LUCENE-2537:

    Attachment: LUCENE-2537.patch

Patch adds the following:
* FSIndexOutput overrides copyBytes to do optimized copies as well (when possible).
* CompoundFileWriter changed to call copyBytes, instead of using an intermediate buffer.
* Some minor typos and code corrections.

*NOTE:* with the changes to CFW, CheckAbort is accessed only after an entire file is written.
The estimated amount of work is the length copied. This means that if abort() is called, it
might take a tad longer until CFW will detect it.
I don't think it's serious, since (1) abort() is not called often, (2) for small segments
this will probably have no effect (OneMerge was accessed roughly every ~2MB copied) and (3)
as reported, the optimized copy is faster when using FileChannel, therefore the time that
passes between checks may not be that long.
And there's a (4) -- for really large segments, the amount of work done to merge them is far
larger than copying them into the CFS. Therefore the chances that abort() will be called during
that process is relatively small ...

All tests pass.

> FSDirectory.copy() impl is unsafe
> ---------------------------------
>                 Key: LUCENE-2537
>                 URL:
>             Project: Lucene - Java
>          Issue Type: Bug
>          Components: Store
>            Reporter: Shai Erera
>            Assignee: Shai Erera
>             Fix For: 3.1, 4.0
>         Attachments:, LUCENE-2537.patch, LUCENE-2537.patch
> There are a couple of issues with it:
> # FileChannel.transferFrom documents that it may not copy the number of bytes requested,
however we don't check the return value. So need to fix the code to read in a loop until all
bytes were copied..
> # When calling addIndexes() w/ very large segments (few hundred MBs in size), I ran into
the following exception (Java 1.6 -- Java 1.5's exception was cryptic):
> {code}
> Exception in thread "main" Map failed
>     at
>     at
>     at
>     at
>     at org.apache.lucene.index.IndexWriter.addIndexes(
> Caused by: java.lang.OutOfMemoryError: Map failed
>     at Method)
>     at
>     ... 7 more
> {code}
> I changed the impl to something like this:
> {code}
> long numWritten = 0;
> long numToWrite = input.size();
> long bufSize = 1 << 26;
> while (numWritten < numToWrite) {
>   numWritten += output.transferFrom(input, numWritten, bufSize);
> }
> {code}
> And the code successfully adds the indexes. This code uses chunks of 64MB, however that
might be too large for some applications, so we definitely need a smaller one. The question
is how small so that performance won't be affected, and it'd be great if we can let it be
configurable, however since that API is called by other API, such as addIndexes, not sure
it's easily controllable.
> Also, I read somewhere (can't remember now where) that on Linux the native impl is better
and does copy in chunks. So perhaps we should make a Linux specific impl?

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message