Interestingly enough, to clean our topologies' code from its use of
Storm shaded classes, I removed storm-shaded-deps.jar from our
dependencies... only to find that in such case we have a compilation
error:
The type org.apache.storm.thrift.TException cannot be resolved. It is
indirectly referenced from required type
org.apache.storm.StormSubmitter
from this type of code (at the last line)
// ----------------------------
// Submit Topology
String displayName =
TopologyHelper.getTopologyDisplayName(topologyShortName,
kafkaBrokersHosts);
StormSubmitter.submitTopology(displayName, topologyConfig,
builder.createTopology());
}
I would have excepted such indirect dependency to be part of
non-shared storm JARs... am I missing something?
Note: in the meantime, let me re-introduce this dependency and use
"find in code" to fix uses of shaded classes..
Thanks,
Alexandre
Le sam. 4 nov. 2023 à 20:57, Alexandre Vermeerbergen
<avermeerber...@gmail.com> a écrit :
>
> Yes another regression I found with my topologies (built with 2.5.0
> dependencies) running to Storm 2.6.0 SNAPSHOT 2023/11/02: I have seen
> several exceptions like this one:
>
> 2023-11-04 18:43:31.696 o.a.s.e.e.ReportError
> Thread-16-alertsHistoryUMToKafka-executor[27, 27] [ERROR] Error
> java.lang.RuntimeException: java.lang.NoClassDefFoundError:
> org.apache.storm.shade.org.json.simple.JSONObject
> at org.apache.storm.utils.Utils$1.run(Utils.java:413)
> ~[storm-client-2.6.0-SNAPSHOT.jar:2.6.0-SNAPSHOT]
> at java.lang.Thread.run(Thread.java:857) [?:?]
> Caused by: java.lang.NoClassDefFoundError:
> org.apache.storm.shade.org.json.simple.JSONObject
> at
> com.acme.storm.evaluator.bolt.AlertHistoryTupleToKafkaMapper.getMessageFromTuple(AlertHistoryTupleToKafkaMapper.java:25)
> ~[stormjar.jar:?]
> at
> com.acme.storm.evaluator.bolt.AlertHistoryTupleToKafkaMapper.getMessageFromTuple(AlertHistoryTupleToKafkaMapper.java:11)
> ~[stormjar.jar:?]
> at org.apache.storm.kafka.bolt.KafkaBolt.process(KafkaBolt.java:177)
> ~[stormjar.jar:?]
>
> and here's the related source code:
>
> package com.acme.storm.evaluator.bolt;
>
> import org.apache.storm.kafka.bolt.mapper.TupleToKafkaMapper;
> import org.apache.storm.shade.org.json.simple.JSONObject;
> import org.apache.storm.tuple.Tuple;
> import org.slf4j.Logger;
> import org.slf4j.LoggerFactory;
>
> import com.acme.storm.util.TupleAlertHistoryFields;
>
> public class AlertHistoryTupleToKafkaMapper implements
> TupleToKafkaMapper<String, String> {
> private static final long serialVersionUID = 5668103889461721134L;
>
> public static final Logger logger =
> LoggerFactory.getLogger(AlertHistoryTupleToKafkaMapper.class);
>
> @Override
> public String getKeyFromTuple(Tuple tuple) {
> return
> tuple.getStringByField(TupleAlertHistoryFields.SERVICE_DEFINITION_NAME);
> }
>
> @SuppressWarnings("unchecked")
> @Override
> public String getMessageFromTuple(Tuple tuple)
> {
> JSONObject tupleJson = new JSONObject();
>
> tuple.getFields().forEach(f -> {
> tupleJson.put(f, tuple.getValueByField(f));
> });
>
> String jsonString = tupleJson.toJSONString();
>
> return jsonString;
> }
> }
>
> line #25 is this one:
>
> JSONObject tupleJson = new JSONObject();
>
> so, from what I understand, Storm 2.6.0 no longer exposes some shaded
> classes such as:
>
> org.apache.storm.shade.org.json.simple.JSONObject;
>
> Interestingly, it looks similar to this issue:
> https://stackoverflow.com/questions/33377291/java-lang-noclassdeffounderror-org-json-simple-jsonvalue-at-storm-kafka-dynamic
>
> Whatever, I guess I can only blame our own code for using Storm's
> shaded classes, right ?
> Note it's been running as is for long time and not breaking since many
> upgrades, of course it's not an excuse...
>
> I'll fix our code & continue checking if I see other impacts of 2.5.0
> -> 2.6.0 upgrade...
>
> Thanks,
> Alexandre
>
> Le sam. 4 nov. 2023 à 10:36, Alexandre Vermeerbergen
> <avermeerber...@gmail.com> a écrit :
> >
> > Just did a 'git push' with that trivial fix... is that enough?
> >
> > Le sam. 4 nov. 2023 à 09:34, Richard Zowalla <rich...@zowalla.com> a écrit :
> > >
> > > You can do as you like ;-)
> > >
> > > Am 4. November 2023 08:54:48 MEZ schrieb Alexandre Vermeerbergen
> > > <avermeerber...@gmail.com>:
> > > >I mean :
> > > >
> > > >Can I keep using my
> > > >'STORM-3958-Capacity-to-set-St
> > > >orm-UI's-title-in-conf/storm.yaml'
> > > >branch to push this fix, or should I create a new one?
> > > >
> > > >Le sam. 4 nov. 2023 à 08:53, Alexandre Vermeerbergen
> > > ><avermeerber...@gmail.com> a écrit :
> > > >>
> > > >> Can you keep using my
> > > >> 'STORM-3958-Capacity-to-set-Storm-UI's-title-in-conf/storm.yaml'
> > > >> branch to push this fix, or should I create a new one?
> > > >>
> > > >> Le sam. 4 nov. 2023 à 08:14, Richard Zowalla <rich...@zowalla.com> a
> > > >> écrit :
> > > >> >
> > > >> > Just push a fix for it (referencing the original issue) ;-) - we
> > > >> > hadn't a release in between, so not a big deal.
> > > >> >
> > > >> > Am 4. November 2023 08:01:54 MEZ schrieb Alexandre Vermeerbergen
> > > >> > <avermeerber...@gmail.com>:
> > > >> > >Temporary workaround for the Storm UI title's regression: run this
> > > >> > >command on the title-template.html file (in
> > > >> > >storm-stable/public/templates directory) after having extracted the
> > > >> > >binary archive:
> > > >> > >
> > > >> > >sed -i 's/{{ui\.title}}/{{title}}/' title-template.html
> > > >> > >
> > > >> > >Again sorry for having introduced this regression.
> > > >> > >
> > > >> > >Alexandre
> > > >> > >
> > > >> > >Le sam. 4 nov. 2023 à 07:52, Alexandre Vermeerbergen
> > > >> > ><avermeerber...@gmail.com> a écrit :
> > > >> > >>
> > > >> > >> Updates regarding my test with Storm 2.6.0 Snapshot 2023/11/02:
> > > >> > >> 1. I have modified my KO bolts using this pattern in their
> > > >> > >> prepare() method:
> > > >> > >>
> > > >> > >> long cacheMaxSize =
> > > >> > >> Long.parseLong(stormConf.get(ConfKeys.SVC_DEF_CACHE_SIZE).toString());
> > > >> > >>
> > > >> > >> instead of:
> > > >> > >>
> > > >> > >> long cacheMaxSize = (Long)
> > > >> > >> stormConf.get(ConfKeys.SVC_DEF_CACHE_SIZE);
> > > >> > >>
> > > >> > >> 2. I have found the cause of the regression in loss of 'Storm UI'
> > > >> > >> title at the top left-hand side of all Storm UI's pages.
> > > >> > >> This is a mistake of mine in this PR:
> > > >> > >>
> > > >> > >> https://github.com/apache/storm/pull/3580/commits/0320ee6f2d367086b973b504a976ae6aff853ccd
> > > >> > >> More precisely, at line 19 of this file:
> > > >> > >> https://github.com/apache/storm/pull/3580/commits/0320ee6f2d367086b973b504a976ae6aff853ccd#diff-6d3a4162a8af826a25a2b8cd6763c08e9bc5165b736b45890b84fa50597d7481
> > > >> > >> we must have:
> > > >> > >> <h1><a href="/">{{title}}</a></h1>
> > > >> > >>
> > > >> > >> instead of:
> > > >> > >>
> > > >> > >> <h1><a href="/">{{ui.title}}</a></h1>
> > > >> > >>
> > > >> > >> I think I missed this during my tests because of browser's
> > > >> > >> cache
> > > >> > >> (bad excuse, I know...): probably I tested by-patching my deployed
> > > >> > >> Storm cluster, and I finally forgot to checkout this latest
> > > >> > >> change of
> > > >> > >> my delivery.
> > > >> > >>
> > > >> > >> I had to use "private navigation" to make sure this time it
> > > >> > >> fixes this issue
> > > >> > >>
> > > >> > >> What would be the quickest way for this very simple fix to be
> > > >> > >> delivered ?
> > > >> > >>
> > > >> > >> Thanks,
> > > >> > >> Alexandre
> > > >> > >>
> > > >> > >> Le ven. 3 nov. 2023 à 20:57, Alexandre Vermeerbergen
> > > >> > >> <avermeerber...@gmail.com> a écrit :
> > > >> > >> >
> > > >> > >> > Thanks Richard for the analysis.
> > > >> > >> >
> > > >> > >> > My thought on this are very simple: let me fix my own old code,
> > > >> > >> > after
> > > >> > >> > all this was impacting only one of my topologies, the other one
> > > >> > >> > were
> > > >> > >> > using more careful Integer.parseInt or Integer.parseLong to
> > > >> > >> > avoid
> > > >> > >> > making assumption on the way the values in the config Map were
> > > >> > >> > serialized.
> > > >> > >> >
> > > >> > >> > My other concern with 2.6.0 Snapshot is that in Storm UI, the
> > > >> > >> > "Storm
> > > >> > >> > UI" title at the top left hand side of each page is missing.
> > > >> > >> > This may
> > > >> > >> > be the result of one of my own PR, I'll have to check why...
> > > >> > >> >
> > > >> > >> > Alex
> > > >> > >> >
> > > >> > >> > Le ven. 3 nov. 2023 à 20:23, Richard Zowalla <r...@apache.org>
> > > >> > >> > a écrit :
> > > >> > >> > >
> > > >> > >> > > Hi,
> > > >> > >> > >
> > > >> > >> > > thanks for the feedback.
> > > >> > >> > >
> > > >> > >> > > The main difference to 2.5.0 is, that we switched the JSON
> > > >> > >> > > implementation in Utils, which relied on a
> > > >> > >> > > super old JSON library. What is happening here is, that the
> > > >> > >> > > new JSON
> > > >> > >> > > library (net.minidev) behaves differently.
> > > >> > >> > >
> > > >> > >> > > We can see it by writing some simple code without Storm (The
> > > >> > >> > > related
> > > >> > >> > > code line with the parsing is taken from Storm Utils)
> > > >> > >> > >
> > > >> > >> > >
> > > >> > >> > > Map<String, Long> conf = new HashMap<>();
> > > >> > >> > > conf.put("test", 100000000L);
> > > >> > >> > >
> > > >> > >> > > Map<String, Object> deserTopoConf = (Map<String, Object>)
> > > >> > >> > > JSONValue.parseWithException(JSONValue.toJSONString(conf));
> > > >> > >> > >
> > > >> > >> > > for(Map.Entry<String, Object> e :
> > > >> > >> > > deserTopoConf.entrySet()) {
> > > >> > >> > >
> > > >> > >> > > System.out.println(e.getValue().getClass().getSimpleName());
> > > >> > >> > > }
> > > >> > >> > >
> > > >> > >> > > This will print "Integer". Using "10000000000000000L" as a
> > > >> > >> > > value will
> > > >> > >> > > result in "Long".
> > > >> > >> > >
> > > >> > >> > > The main issue derives from the fact, that we serialize the
> > > >> > >> > > config map
> > > >> > >> > > into a JSON String and parse the result.
> > > >> > >> > > Based on the String representation, "100000000L" will still
> > > >> > >> > > fit into an
> > > >> > >> > > Integer, so the parser will not use Long ;-)
> > > >> > >> > >
> > > >> > >> > > If the number is big enough, the parser will switch to a Long.
> > > >> > >> > > The old library (> 10 years old now):
> > > >> > >> > >
> > > >> > >> > > deserTopoConf = (Map<String, Object>)
> > > >> > >> > > org.json.simple.JSONValue.parseWithException(org.json.simple.JSONValue.
> > > >> > >> > > toJSONString(conf));
> > > >> > >> > >
> > > >> > >> > > for(Map.Entry<String, Object> e :
> > > >> > >> > > deserTopoConf.entrySet()) {
> > > >> > >> > >
> > > >> > >> > > System.out.println(e.getValue().getClass().getSimpleName());
> > > >> > >> > > }
> > > >> > >> > >
> > > >> > >> > > This will print "Long".
> > > >> > >> > >
> > > >> > >> > > The difference between these libraries is, that the super old
> > > >> > >> > > org.json.simple just always returns Long as the type.
> > > >> > >> > >
> > > >> > >> > > If can eventually tune the new JSON parser to something like:
> > > >> > >> > >
> > > >> > >> > > deserTopoConf = (Map<String, Object>) new
> > > >> > >> > > JSONParser(JSONParser.BIG_DIGIT_UNRESTRICTED).parse(JSONValue.toJSONStr
> > > >> > >> > > ing(conf), JSONValue.defaultReader.DEFAULT);
> > > >> > >> > >
> > > >> > >> > > This will result in "Long" but might have other side-effects.
> > > >> > >> > >
> > > >> > >> > > From my point of view, the newer JSON library works as
> > > >> > >> > > expected
> > > >> > >> > > (although it might break backward compatibility).
> > > >> > >> > > Unchecked casts are always evil, imho. In your specific case,
> > > >> > >> > > it might
> > > >> > >> > > be more save to cast to "Number" and than invoke
> > > >> > >> > > "longValue()" instead.
> > > >> > >> > >
> > > >> > >> > > Nevertheless, we might being able to restore the old
> > > >> > >> > > behaviour by
> > > >> > >> > > tuning the JSONParser configuration.
> > > >> > >> > >
> > > >> > >> > > Thoughts?
> > > >> > >> > >
> > > >> > >> > > Gruß
> > > >> > >> > > Richard
> > > >> > >> > >
> > > >> > >> > >
> > > >> > >> > >
> > > >> > >> > >
> > > >> > >> > > Am Freitag, dem 03.11.2023 um 18:32 +0100 schrieb Alexandre
> > > >> > >> > > Vermeerbergen:
> > > >> > >> > > > On a side note, the Long value that "MyKOBolt" is set by a
> > > >> > >> > > > DefaultTopologyConfig to a value with is a java.lang.Long
> > > >> > >> > > > instance...
> > > >> > >> > > > hence it's surprising to see it internally converted into a
> > > >> > >> > > > java.lang.Integer in the Map passed to prepare() method...
> > > >> > >> > > >
> > > >> > >> > > >
> > > >> > >> > > > Le ven. 3 nov. 2023 à 18:27, Alexandre Vermeerbergen
> > > >> > >> > > > <avermeerber...@gmail.com> a écrit :
> > > >> > >> > > > >
> > > >> > >> > > > > Hello,
> > > >> > >> > > > >
> > > >> > >> > > > > I found a breaking change using Apache Storm 2.6.0
> > > >> > >> > > > > snapshot build
> > > >> > >> > > > > dated 20231103 and without changing my topologies' code
> > > >> > >> > > > > (keeping
> > > >> > >> > > > > the
> > > >> > >> > > > > same big jars).
> > > >> > >> > > > >
> > > >> > >> > > > > This seems to be tied to a datatype change in Map
> > > >> > >> > > > > argument of
> > > >> > >> > > > > prepare() method of Bolt.
> > > >> > >> > > > >
> > > >> > >> > > > > For example, this Bolt is KO at runtime:
> > > >> > >> > > > >
> > > >> > >> > > > > public class MyKOBolt extends BaseRichBolt {
> > > >> > >> > > > >
> > > >> > >> > > > > @Override
> > > >> > >> > > > > public void prepare(Map stormConf, TopologyContext
> > > >> > >> > > > > context,
> > > >> > >> > > > > OutputCollector collector) {
> > > >> > >> > > > > cacheMaxSize = (Long)
> > > >> > >> > > > > stormConf.get(ConfKeys.SVC_DEF_CACHE_SIZE);
> > > >> > >> > > > > cacheTtlMin = (Long)
> > > >> > >> > > > > stormConf.get(ConfKeys.SVC_DEF_CACHE_TTL_MIN);
> > > >> > >> > > > >
> > > >> > >> > > > > as it throws this error:
> > > >> > >> > > > > 2023-11-03 16:42:20.230 o.a.s.e.b.BoltExecutor
> > > >> > >> > > > > Thread-14-__system-executor[-1, -1] [INFO] Preparing bolt
> > > >> > >> > > > > __system:[-1]
> > > >> > >> > > > > 2023-11-03 16:42:20.193 o.a.s.u.Utils
> > > >> > >> > > > > Thread-15-evaluateTriggers-executor[11, 11] [ERROR] Async
> > > >> > >> > > > > loop
> > > >> > >> > > > > died!
> > > >> > >> > > > > java.lang.ClassCastException: java.lang.Integer
> > > >> > >> > > > > incompatible with
> > > >> > >> > > > > java.lang.Long
> > > >> > >> > > > > at com.acme.storm.alerting.MyKOBolt
> > > >> > >> > > > > .prepare(EvaluationBolt.java:71)
> > > >> > >> > > > > ~[stormjar.jar:?]
> > > >> > >> > > > > at
> > > >> > >> > > > > org.apache.storm.executor.bolt.BoltExecutor.init(BoltExecutor.java:
> > > >> > >> > > > > 128)
> > > >> > >> > > > > ~[storm-client-2.6.0-SNAPSHOT.jar:2.6.0-SNAPSHOT]
> > > >> > >> > > > > at
> > > >> > >> > > > > org.apache.storm.executor.bolt.BoltExecutor.call(BoltExecutor.java:
> > > >> > >> > > > > 138)
> > > >> > >> > > > > ~[storm-client-2.6.0-SNAPSHOT.jar:2.6.0-SNAPSHOT]
> > > >> > >> > > > > at
> > > >> > >> > > > > org.apache.storm.executor.bolt.BoltExecutor.call(BoltExecutor.java:
> > > >> > >> > > > > 54)
> > > >> > >> > > > > ~[storm-client-2.6.0-SNAPSHOT.jar:2.6.0-SNAPSHOT]
> > > >> > >> > > > > at org.apache.storm.utils.Utils$1.run(Utils.java:393)
> > > >> > >> > > > > [storm-client-2.6.0-SNAPSHOT.jar:2.6.0-SNAPSHOT]
> > > >> > >> > > > > at java.lang.Thread.run(Thread.java:857) [?:?]
> > > >> > >> > > > >
> > > >> > >> > > > > whereas this bolt is fine with Storm 2.6.0:
> > > >> > >> > > > >
> > > >> > >> > > > > public class MyOKBolt extends BaseBasicBolt {
> > > >> > >> > > > >
> > > >> > >> > > > > @Override
> > > >> > >> > > > > public void prepare(final Map stormConf, final
> > > >> > >> > > > > TopologyContext
> > > >> > >> > > > > context) {
> > > >> > >> > > > > super.prepare(stormConf, context);
> > > >> > >> > > > >
> > > >> > >> > > > > this.redisPort =
> > > >> > >> > > > > Integer.parseInt((String)
> > > >> > >> > > > > stormConf.get(ConfKeys.REDIS_PORT));
> > > >> > >> > > > >
> > > >> > >> > > > > That said, I can modify the code of "MyKOBolt" to use same
> > > >> > >> > > > > Integer.parseInt or Integer.parseLong trick, but it's
> > > >> > >> > > > > the first
> > > >> > >> > > > > time
> > > >> > >> > > > > in my long history of upgrades that I have seen such
> > > >> > >> > > > > runtime
> > > >> > >> > > > > incompatibility.
> > > >> > >> > > > >
> > > >> > >> > > > > Maybe "MyKOBolt" was badly written since the beginning
> > > >> > >> > > > > and I have
> > > >> > >> > > > > just
> > > >> > >> > > > > hit the punishment for it: is there a documentation which
> > > >> > >> > > > > clarifies
> > > >> > >> > > > > datatypes of the Map argument of prepare() method?
> > > >> > >> > > > >
> > > >> > >> > > > > Also worth noting: "MyKOBolt" derives from BaseRichBolt ,
> > > >> > >> > > > > while
> > > >> > >> > > > > "MyOKBolt" derives from BaseBasicBolt => could this have
> > > >> > >> > > > > any impact
> > > >> > >> > > > > on
> > > >> > >> > > > > this finding ?
> > > >> > >> > > > >
> > > >> > >> > > > > Last, I'm running this on Redhat Linux 8 and IBM Semeru
> > > >> > >> > > > > JDK
> > > >> > >> > > > > 17.0.8.1.
> > > >> > >> > > > >
> > > >> > >> > > > > Kind regards,
> > > >> > >> > > > > Alexandre
> > > >> > >> > > > >
> > > >> > >> > > > > Le ven. 3 nov. 2023 à 15:16, Julien Nioche
> > > >> > >> > > > > <lists.digitalpeb...@gmail.com> a écrit :
> > > >> > >> > > > > >
> > > >> > >> > > > > > Thanks Richard.
> > > >> > >> > > > > >
> > > >> > >> > > > > > Tried the latest snapshot with StormCrawler both in
> > > >> > >> > > > > > local and
> > > >> > >> > > > > > deployed mode
> > > >> > >> > > > > > and did not find any issues.
> > > >> > >> > > > > > Will try it on a topology generating WARC files next
> > > >> > >> > > > > > week to
> > > >> > >> > > > > > check that the
> > > >> > >> > > > > > dependency changes on Hadoop have not broken anything.
> > > >> > >> > > > > >
> > > >> > >> > > > > > Have a good week end
> > > >> > >> > > > > >
> > > >> > >> > > > > > Julien
> > > >> > >> > > > > >
> > > >> > >> > > > > > On Thu, 2 Nov 2023 at 19:25, Richard Zowalla
> > > >> > >> > > > > > <r...@apache.org>
> > > >> > >> > > > > > wrote:
> > > >> > >> > > > > >
> > > >> > >> > > > > > > Short update. Sorted out an issue with the tar.gz/zip
> > > >> > >> > > > > > > with
> > > >> > >> > > > > > > Julien today
> > > >> > >> > > > > > > and re-uploaded them to the nightlies area.
> > > >> > >> > > > > > >
> > > >> > >> > > > > > > This new bundle works as expected in my deployment
> > > >> > >> > > > > > > but happy to
> > > >> > >> > > > > > > receive
> > > >> > >> > > > > > > additional feedback before getting up a first release
> > > >> > >> > > > > > > candidate
> > > >> > >> > > > > > > ;-)
> > > >> > >> > > > > > >
> > > >> > >> > > > > > > Am Montag, dem 30.10.2023 um 08:21 +0100 schrieb
> > > >> > >> > > > > > > Richard
> > > >> > >> > > > > > > Zowalla:
> > > >> > >> > > > > > > > Hi Alexandre,
> > > >> > >> > > > > > > >
> > > >> > >> > > > > > > > we are not in a hurry here :) - take as much as
> > > >> > >> > > > > > > > time you
> > > >> > >> > > > > > > > need.
> > > >> > >> > > > > > > >
> > > >> > >> > > > > > > > Gruß
> > > >> > >> > > > > > > > Richard
> > > >> > >> > > > > > > >
> > > >> > >> > > > > > > >
> > > >> > >> > > > > > > > Am Montag, dem 30.10.2023 um 07:50 +0100 schrieb
> > > >> > >> > > > > > > > Alexandre
> > > >> > >> > > > > > > > Vermeerbergen:
> > > >> > >> > > > > > > > > Hello Richard,
> > > >> > >> > > > > > > > >
> > > >> > >> > > > > > > > > Okay, I'm more than happy to do that with my
> > > >> > >> > > > > > > > > pre-production
> > > >> > >> > > > > > > > > cluster
> > > >> > >> > > > > > > > > (~10 topologies) using the binary artifacts.
> > > >> > >> > > > > > > > > Would it be OK if I can use up to end of this
> > > >> > >> > > > > > > > > week so that
> > > >> > >> > > > > > > > > I'll be
> > > >> > >> > > > > > > > > able to have enough time to check all potential
> > > >> > >> > > > > > > > > issues that
> > > >> > >> > > > > > > > > this
> > > >> > >> > > > > > > > > upgrade bring?
> > > >> > >> > > > > > > > >
> > > >> > >> > > > > > > > > Thanks,
> > > >> > >> > > > > > > > > Alexandre
> > > >> > >> > > > > > > > >
> > > >> > >> > > > > > > > > Le lun. 30 oct. 2023 à 07:35, Richard Zowalla
> > > >> > >> > > > > > > > > <r...@apache.org> a
> > > >> > >> > > > > > > > > écrit :
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > > Hi all,
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > > before starting a release of 2.6.0-SNAPSHOT, I
> > > >> > >> > > > > > > > > > would like
> > > >> > >> > > > > > > > > > to
> > > >> > >> > > > > > > > > > receive
> > > >> > >> > > > > > > > > > some feedback on the current SNAPSHOT build.
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > > I just uploaded a 2.6.0-SNAPSHOT of Storm build
> > > >> > >> > > > > > > > > > from
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > https://github.com/apache/storm/commit/8f883086032669a8f04b09a3b312d60af5b44533
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > > It is available via the ASF Snapshot repository:
> > > >> > >> > > > > > > > > > https://repository.apache.org/content/repositories/snapshots/
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > > You can consume by adding
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > > <repository>
> > > >> > >> > > > > > > > > > <id>apache.snapshots</id>
> > > >> > >> > > > > > > > > > <name>Apache Snapshot Repository</name>
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > > <url>https://repository.apache.org/snapshots</url>
> > > >> > >> > > > > > > > > > <releases>
> > > >> > >> > > > > > > > > > <enabled>false</enabled>
> > > >> > >> > > > > > > > > > </releases>
> > > >> > >> > > > > > > > > > </repository>
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > > to your project pom. As we do not deploy
> > > >> > >> > > > > > > > > > SNAPSHOTS
> > > >> > >> > > > > > > > > > automatically,
> > > >> > >> > > > > > > > > > it
> > > >> > >> > > > > > > > > > should be easy to just consume the latest
> > > >> > >> > > > > > > > > > SNAPSHOT.
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > > The packaged binaries are available at
> > > >> > >> > > > > > > > > > nightlies.apache.org:
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > https://nightlies.apache.org/storm/2.6.0-SNAPSHOT/8f883086032669a8f04b09a3b312d60af5b44533/
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > > If you have some minutes left: Please test and
> > > >> > >> > > > > > > > > > report any
> > > >> > >> > > > > > > > > > issues
> > > >> > >> > > > > > > > > > with
> > > >> > >> > > > > > > > > > this binaries, so we can fix before attempting
> > > >> > >> > > > > > > > > > to
> > > >> > >> > > > > > > > > > release.
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > > The most significant changes are in the are of
> > > >> > >> > > > > > > > > > hadoop/hbase/hdfs
> > > >> > >> > > > > > > > > > as
> > > >> > >> > > > > > > > > > we
> > > >> > >> > > > > > > > > > upgraded from 2.x to 3.x - our own test
> > > >> > >> > > > > > > > > > coverage within
> > > >> > >> > > > > > > > > > the build
> > > >> > >> > > > > > > > > > looks
> > > >> > >> > > > > > > > > > good but would be nice to get some real world
> > > >> > >> > > > > > > > > > use-case
> > > >> > >> > > > > > > > > > feedback
> > > >> > >> > > > > > > > > > on
> > > >> > >> > > > > > > > > > this. In addition, we had quite a lof of 3rd
> > > >> > >> > > > > > > > > > party
> > > >> > >> > > > > > > > > > dependency
> > > >> > >> > > > > > > > > > upgrades.
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > > In addition, it contains the pruning of
> > > >> > >> > > > > > > > > > external modules
> > > >> > >> > > > > > > > > > as
> > > >> > >> > > > > > > > > > listed
> > > >> > >> > > > > > > > > > in
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > > https://issues.apache.org/jira/browse/STORM-3988
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > > A summaryin Jira is here:
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > https://issues.apache.org/jira/secure/ReleaseNote.jspa?projectId=12314820&version=12353484
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > > Gruß
> > > >> > >> > > > > > > > > > Richard
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > > > >
> > > >> > >> > > > > > > >
> > > >> > >> > > > > > >
> > > >> > >> > > > > > >
> > > >> > >> > > > > >
> > > >> > >> > > > > > --
> > > >> > >> > > > > >
> > > >> > >> > > > > > *Open Source Solutions for Text Engineering*
> > > >> > >> > > > > >
> > > >> > >> > > > > > http://www.digitalpebble.com
> > > >> > >> > > > > > http://digitalpebble.blogspot.com/
> > > >> > >> > > > > > #digitalpebble <http://twitter.com/digitalpebble>
> > > >> > >> > >
