Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Plain View
Drill, mail # dev - How to enable Drill log to analyze query execution


+
Jinfeng Ni 2013-09-14, 06:29
+
Timothy Chen 2013-09-14, 15:19
+
Jinfeng Ni 2013-09-14, 22:37
Copy link to this message
-
Re: How to enable Drill log to analyze query execution
Steven Phillips 2013-09-14, 23:05
You are thinking about it backwards. You don't configure Lilith to connect
to the remote service. Lilith listens for remote services to it. You have
to configure the logback.xml file used by the drillbit to point the host
that is running lilith:

<appender name="SOCKET"

class="de.huxhorn.lilith.logback.appender.ClassicMultiplexSocketAppender">
    <Compressing>true</Compressing>
    <ReconnectionDelay>10000</ReconnectionDelay>
    <IncludeCallerData>true</IncludeCallerData>
    <RemoteHosts>localhost</RemoteHosts>
  </appender>

replace localhost with the hostname or ip address where lilith is run.

You can use this, for example, to have drillbits running on multiple
machines, and have the log messages sent back to your laptop.
On Sat, Sep 14, 2013 at 3:37 PM, Jinfeng Ni <[EMAIL PROTECTED]> wrote:

> Hi Tim,
>
> Thanks for your tips.  Regarding Lilth, my questions is whether Lilth and
> Drill have to be on the same machine, or they can run on different
> machines.   i.e, Drill on Ubuntu box, and  Lilth on Windows 7 machine.  I
> run Lilth on Windows box, which simply shows "No active connections", and I
> could not see way in Lilth to specify how to connect to the Drill server on
> Ubuntu box.
>
> Thanks!
>
>
>
>
>
>
> On Sat, Sep 14, 2013 at 8:19 AM, Timothy Chen <[EMAIL PROTECTED]> wrote:
>
> > Hi Jinfeng,
> >
> > Once you get Lilth installed (which is simply downloading it and running
> > it) and ran, it will be listening to the socket that Drill opens to write
> > to.
> >
> > If you then launches sqline and run queries, there should be logs showing
> > up on Lilth automatically.
> >
> > As for Eclipse problem, I never used Eclipse with Drill so hopefully
> > someone else can answer that question for you.
> >
> > Tim
> >
> >
> > On Fri, Sep 13, 2013 at 11:29 PM, Jinfeng Ni <[EMAIL PROTECTED]>
> > wrote:
> >
> > > Hi all,
> > >
> > > Hope you don't mind if I ask couple of "beginner" questions, as Drill
> is
> > a
> > > brand new thing to me currently.
> > >
> > > 1. Enable Drill log.
> > > I tried to follow the instruction at
> > >
> > >
> >
> https://cwiki.apache.org/confluence/display/DRILL/How+To+Run+Drill+with+Sample+Data
> > >
> > > At the bottom, it suggested use Lilith to view / analyze the log from
> > > Drill. I went through the links there, still have no idea how to set
> up .
> > >
> > >  Q1 : Do I have to install Lilith on the same machine where Drill is
> > > running?  What kind of connection will Lilith use to "fetch" log data
> > from
> > > Drill?
> > >
> > > I have a Ubuntu 12.0.4 box with Drill, and Lilith installed on a
> Windows
> > > desktop. Can this kind of configuration work?
> > >
> > >   Q2:  Is there any setup so that the "logback socket appender" for
> Drill
> > > will be enabled?
> > >
> > > 2.  Eclipse IDE
> > >
> > > I followed instructions
> > >
> > >
> >
> https://cwiki.apache.org/confluence/display/DRILL/Sources+and+Setting+Up+Development+Environment
> > > ,
> > > which probably is a bit old (Dated Feb, 2013).
> > >
> > > After import the source code into Eclipse IDE running on Windows
> Desktop,
> > > it shows many packages are missing, and compiler errors.
> > >
> > > On Ubuntu box, I noticed the directory .m2/repository contains lots of
> > > dependency .jar, and the $DRILL_HOME/.classpath file will add those
> .jar
> > to
> > > classpath.
> > >
> > > On Windows Eclipse,  I also saw directory .m2/repository.  However, the
> > > Eclipse project seems not know how to locate those dependency .jar.
> > >
> > > My question is :  Do I have to "manually" add those .jar to the
> project's
> > > build path, to fix the compiler error from Eclipse?
> > >
> > > PS: In Eclipse, "Maven install" failed with error:
> > >
> > >
> >
> testMultipleRowGroups(org.apache.drill.exec.store.ParquetRecordReaderTest)
> > >  Time elapsed: 0.138 sec  <<< ERROR!
> > > java.io.IOException: File already exists:/tmp/test.parquet
> > >
> > >
> > > Thanks a lot for any help!
> > >
> >
>
+
Jinfeng Ni 2013-09-14, 23:26