Hi Cheng,
I tried both these patches, and seems still not resolve my issue. And I
found the most time is spend on this line in newParquet.scala:
ParquetFileReader.readAllFootersInParallel(
sparkContext.hadoopConfiguration, seqAsJavaList(leaves), taskSideMetaData)
Which need read all the files under the Parquet folder, while our Parquet
folder has a lot of Parquet files (near 2000), read one file need about 2
seconds, so it become very slow ... And the PR 5231 did not skip this steps
so it not resolve my issue.
As our Parquet files are generated by a Spark job, so the number of
.parquet files is same with the number of tasks, that is why we have so
many files. But these files actually have the same schema. Is there any way
to merge these files into one, or avoid scan each of them?
On Sat, Apr 4, 2015 at 9:47 PM, Cheng Lian lian.cs@gmail.com wrote:
Hey Xudong,
We had been digging this issue for a while, and believe PR 5339
http://github.com/apache/spark/pull/5339 and PR 5334
http://github.com/apache/spark/pull/5339 should fix this issue.
There two problems:
1. Normally we cache Parquet table metadata for better performance, but
when converting Hive metastore Hive tables, the cache is not used. Thus
heavy operations like schema discovery is done every time a metastore
Parquet table is converted.
2. With Parquet task side metadata reading (which is turned on by
default), we can actually skip the row group information in the footer.
However, we accidentally called a Parquet function which doesn't skip row
group information.
For your question about schema merging, Parquet allows different
part-files have different but compatible schemas. For example,
part-1.parquet has columns a and b, while part-2.parquet may has
columns a and c. In some cases, the summary files (_metadata and
_common_metadata) contains the merged schema (a, b, and c), but it's not
guaranteed. For example, when the user defined metadata stored different
part-files contain different values for the same key, Parquet simply gives
up writing summary files. That's why all part-files must be touched to get
a precise merged schema.
However, in scenarios where a centralized arbitrative schema is available
(e.g. Hive metastore schema, or the schema provided by user via data source
DDL), we don't need to do schema merging on driver side, but defer it to
executor side and each task only needs to reconcile those part-files it
needs to touch. This is also what the Parquet developers did recently for
parquet-hadoop https://github.com/apache/incubator-parquet-mr/pull/45.
Cheng
On 3/31/15 11:49 PM, Zheng, Xudong wrote:
Thanks Cheng!
Set 'spark.sql.parquet.useDataSourceApi' to false resolves my issues,
but the PR 5231 seems not. Not sure any other things I did wrong ...
BTW, actually, we are very interested in the schema merging feature in
Spark 1.3, so both these two solution will disable this feature, right? It
seems that Parquet metadata is store in a file named _metadata in the
Parquet file folder (each folder is a partition as we use partition table),
why we need scan all Parquet part files? Is there any other solutions could
keep schema merging feature at the same time? We are really like this
feature :)
On Tue, Mar 31, 2015 at 3:19 PM, Cheng Lian lian.cs@gmail.com wrote:
Hi Xudong,
This is probably because of Parquet schema merging is turned on by
default. This is generally useful for Parquet files with different but
compatible schemas. But it needs to read metadata from all Parquet
part-files. This can be problematic when reading Parquet files with lots of
part-files, especially when the user doesn't need schema merging.
This issue is tracked by SPARK-6575, and here is a PR for it:
https://github.com/apache/spark/pull/5231. This PR adds a configuration
to disable schema merging by default when doing Hive metastore Parquet
table conversion.
Another workaround is to fallback to the old Parquet code by setting
spark.sql.parquet.useDataSourceApi to false.
Cheng
On 3/31/15 2:47 PM, Zheng, Xudong wrote:
Hi all,
We are using Parquet Hive table, and we are upgrading to Spark 1.3. But
we find that, just a simple COUNT(*) query will much slower (100x) than
Spark 1.2.
I find the most time spent on driver to get HDFS blocks. I find large
amount of get below logs printed:
15/03/30 23:03:43 DEBUG ProtobufRpcEngine: Call: getBlockLocations took
2097ms
15/03/30 23:03:43 DEBUG DFSClient: newInfo = LocatedBlocks{
fileLength=77153436
underConstruction=false
blocks=[LocatedBlock{BP-1236294426-10.152.90.181-1425290838173:blk_1075187948_1448275;
getBlockSize()=77153436; corrupt=false; offset=0;
locs=[10.152.116.172:50010, 10.152.116.169:50010, 10.153.125.184:50010]}]
lastLocatedBlock=LocatedBlock{BP-1236294426-10.152.90.181-1425290838173:blk_1075187948_1448275;
getBlockSize()=77153436; corrupt=false; offset=0;
locs=[10.152.116.169
