Here is the full configuration of Rocks.

image.png

Le mer. 1 avr. 2020 à 10:41, Nicolas Carlot <nicolas.carlot@chronopost.fr> a écrit :
It's not that I cannot turn on compaction.
Compaction works fine.
The issue is with the restoration process of the state store, which tries to compact the store to a single level: db.compactRange(columnFamily, true, 1, 0) before bulk loading the data.
It automatically fails when I'm using FIFO compaction.


Le mer. 1 avr. 2020 à 10:26, Nicolas Carlot <nicolas.carlot@chronopost.fr> a écrit :
My current workaround is to completely delete the state store and rebuild it from scratch.

Le mar. 31 mars 2020 à 21:39, Boyang Chen <reluctanthero104@gmail.com> a écrit :
Thanks Nicolas for the report, so are you suggesting that you couldn't turn
on compactions for the state store? Is there a workaround?

On Tue, Mar 31, 2020 at 9:54 AM Nicolas Carlot <nicolas.carlot@chronopost.fr>
wrote:

> After some more testing and debugging, it seems that it is caused by the
> compaction option I've configured for RocksDB. When removed everything is
> fine...
> The option is as follow:
>
> CompactionOptionsFIFO fifoOptions = new CompactionOptionsFIFO();
> fifoOptions.setMaxTableFilesSize(maxSize);
> fifoOptions.setAllowCompaction(true);
> options.setCompactionOptionsFIFO(fifoOptions);
> options.setCompactionStyle(CompactionStyle.FIFO);
>
> Le mar. 31 mars 2020 à 16:27, Nicolas Carlot <nicolas.carlot@chronopost.fr
> >
> a écrit :
>
> > Hello everyone,
> >
> > I'm currently facing an issue with RocksDb internal compaction process,
> > which occurs when the local state store of several of my KafkaStream
> > applications are being restored. This is sadly a huge concern as it
> > completely discard resiliency over node failure as those often lead to a
> > state store restoration. The only workaround I currently have is to
> delete
> > the local store to restore it from scratch. I'm using version 2.4.1 of
> the
> > Java libraries.
> >
> > The exception thrown by the KStream process is:
> > org.apache.kafka.streams.errors.ProcessorStateException: Error while
> range
> > compacting during restoring  store merge_store
> >         at
> >
> org.apache.kafka.streams.state.internals.RocksDBStore$SingleColumnFamilyAccessor.toggleDbForBulkLoading(RocksDBStore.java:615)
> > ~[kafka-stream-router.jar:?]
> >         at
> >
> org.apache.kafka.streams.state.internals.RocksDBStore.toggleDbForBulkLoading(RocksDBStore.java:398)
> > ~[kafka-stream-router.jar:?]
> >         at
> >
> org.apache.kafka.streams.state.internals.RocksDBStore$RocksDBBatchingRestoreCallback.onRestoreStart(RocksDBStore.java:644)
> > ~[kafka-stream-router.jar:?]
> >         at
> >
> org.apache.kafka.streams.processor.internals.CompositeRestoreListener.onRestoreStart(CompositeRestoreListener.java:59)
> > ~[kafka-stream-router.jar:?]
> >         at
> >
> org.apache.kafka.streams.processor.internals.StateRestorer.restoreStarted(StateRestorer.java:76)
> > ~[kafka-stream-router.jar:?]
> >         at
> >
> org.apache.kafka.streams.processor.internals.StoreChangelogReader.startRestoration(StoreChangelogReader.java:211)
> > ~[kafka-stream-router.jar:?]
> >         at
> >
> org.apache.kafka.streams.processor.internals.StoreChangelogReader.initialize(StoreChangelogReader.java:185)
> > ~[kafka-stream-router.jar:?]
> >         at
> >
> org.apache.kafka.streams.processor.internals.StoreChangelogReader.restore(StoreChangelogReader.java:81)
> > ~[kafka-stream-router.jar:?]
> >         at
> >
> org.apache.kafka.streams.processor.internals.TaskManager.updateNewAndRestoringTasks(TaskManager.java:389)
> > ~[kafka-stream-router.jar:?]
> >         at
> >
> org.apache.kafka.streams.processor.internals.StreamThread.runOnce(StreamThread.java:769)
> > ~[kafka-stream-router.jar:?]
> >         at
> >
> org.apache.kafka.streams.processor.internals.StreamThread.runLoop(StreamThread.java:698)
> > ~[kafka-stream-router.jar:?]
> >         at
> >
> org.apache.kafka.streams.processor.internals.StreamThread.run(StreamThread.java:671)
> > [kafka-stream-router.jar:?]
> > Caused by: org.rocksdb.RocksDBException: Target level exceeds number of
> > levels
> >         at org.rocksdb.RocksDB.compactRange(Native Method)
> > ~[kafka-stream-router.jar:?]
> >         at org.rocksdb.RocksDB.compactRange(RocksDB.java:2636)
> > ~[kafka-stream-router.jar:?]
> >         at
> >
> org.apache.kafka.streams.state.internals.RocksDBStore$SingleColumnFamilyAccessor.toggleDbForBulkLoading(RocksDBStore.java:613)
> > ~[kafka-stream-router.jar:?]
> >         ... 11 more
> >
> > Here is the state of partition 0_0 of one of those stores:
> > -rw-r--r-- 1 java j2ee 148210568 31 mars  10:49 000139.sst
> > -rw-r--r-- 1 java j2ee  21620385 31 mars  11:06 000184.sst
> > -rw-r--r-- 1 java j2ee         0 31 mars  11:11 000198.log
> > -rw-r--r-- 1 java j2ee  31602468 31 mars  11:31 000251.sst
> > -rw-r--r-- 1 java j2ee  37856549 31 mars  12:00 000324.sst
> > -rw-r--r-- 1 java j2ee  33498822 31 mars  12:26 000393.sst
> > -rw-r--r-- 1 java j2ee  34368461 31 mars  12:49 000450.sst
> > -rw-r--r-- 1 java j2ee  11371247 31 mars  12:55 000467.sst
> > -rw-r--r-- 1 java j2ee  14356435 31 mars  13:04 000489.sst
> > -rw-r--r-- 1 java j2ee   5858737 31 mars  13:05 000494.sst
> > -rw-r--r-- 1 java j2ee   2545952 31 mars  14:08 000659.sst
> > -rw-r--r-- 1 java j2ee   3187275 31 mars  15:27 000868.sst
> > -rw-r--r-- 1 java j2ee    407017 31 mars  15:34 000885.sst
> > -rw-r--r-- 1 java j2ee    590190 31 mars  15:45 000914.sst
> > -rw-r--r-- 1 java j2ee    154471 31 mars  15:47 000919.sst
> > -rw-r--r-- 1 java j2ee    139838 31 mars  15:49 000924.sst
> > -rw-r--r-- 1 java j2ee     35058 31 mars  15:49 000925.sst
> > -rw-r--r-- 1 java j2ee     33987 31 mars  15:50 000926.sst
> > -rw-r--r-- 1 java j2ee        16 31 mars  11:11 CURRENT
> > -rw-r--r-- 1 java j2ee        37 31 mars  10:33 IDENTITY
> > -rw-r--r-- 1 java j2ee         0 31 mars  10:33 LOCK
> > -rw-r--r-- 1 java j2ee     15340 31 mars  11:11 LOG
> > -rw-r--r-- 1 java j2ee     15046 31 mars  10:33 LOG.old.1585643630145007
> > -rw-r--r-- 1 java j2ee     15290 31 mars  10:33 LOG.old.1585643826265995
> > -rw-r--r-- 1 java j2ee     15384 31 mars  10:37 LOG.old.1585645861692248
> > -rw-r--r-- 1 java j2ee     60767 31 mars  15:55 MANIFEST-000197
> > -rw-r--r-- 1 java j2ee      4857 31 mars  10:37 OPTIONS-000107
> > -rw-r--r-- 1 java j2ee      4857 31 mars  11:11 OPTIONS-000200
> >
> > I can see that Kafka is running RocksDB's compaction with:
> > public void toggleDbForBulkLoading() {
> >             try {
> >                 db.compactRange(columnFamily, true, 1, 0);
> >             } catch (final RocksDBException e) {
> >                 throw new ProcessorStateException("Error while range
> > compacting during restoring  store " + name, e);
> >             }
> >         }
> >
> > Seems related to an issue with RocksDB itself:
> > https://github.com/facebook/rocksdb/issues/2734
> > But i'm not sure of this.
> > Any help would be greatly appreciated here :)
> >
>
>
> --
> *Nicolas Carlot*
>
> Lead dev
> |  | nicolas.carlot@chronopost.fr
>
>
> *Veuillez noter qu'à partir du 20 mai, le siège Chronopost déménage. La
> nouvelle adresse est : 3 boulevard Romain Rolland 75014 Paris*
>
> [image: Logo Chronopost]
> | chronopost.fr <http://www.chronopost.fr/>
> Suivez nous sur Facebook <https://fr-fr.facebook.com/chronopost> et
> Twitter
> <https://twitter.com/chronopost>.
>
> [image: DPD Group]
>


--
Nicolas Carlot
 
Lead dev

Veuillez noter qu'à partir du 20 mai, le siège Chronopost déménage.
La nouvelle adresse est : 3 boulevard Romain Rolland 75014 Paris
 
Logo Chronopost
chronopost.fr
Suivez nous sur Facebook et Twitter.

DPD Group


--
Nicolas Carlot
 
Lead dev
|  | nicolas.carlot@chronopost.fr

Veuillez noter qu'à partir du 20 mai, le siège Chronopost déménage.
La nouvelle adresse est : 3 boulevard Romain Rolland 75014 Paris
 
Logo Chronopost
chronopost.fr
Suivez nous sur Facebook et Twitter.

DPD Group


--
Nicolas Carlot
 
Lead dev
|  | nicolas.carlot@chronopost.fr

Veuillez noter qu'à partir du 20 mai, le siège Chronopost déménage.
La nouvelle adresse est : 3 boulevard Romain Rolland 75014 Paris
 
Logo Chronopost
chronopost.fr
Suivez nous sur Facebook et Twitter.

DPD Group