You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@esme.apache.org by Andy the destroyer <an...@gmail.com> on 2009/08/03 21:12:26 UTC
Stats code
Hi all,
Attached is a patch file containing basic stats gathering code. I introduced
a dependency on a scala-stats project I forked from the twitter guys and
modified. The scala-stats dependency can be downloaded from
http://github.com/andythedestroyer/scala-stats/downloads.
I initially set up 19 counters for:
* liftSessions - ( actually grabbed from SessionWatcher )
* usersLoggedIn ( incremented and decremented through User.logUserIn and
User logUserOut)
* userCount ( User actors started since start/reset )
* groupCount
* messagesCreated
* userMessagesCreated
* interpreterMessagesCreated
* schedulerMessagesCreated
* messagesPulled ( messages pulled from MessagePullActor )
* messagesDelivered
* messagesDeliveredTrackReason
* messagesDeliveredDirectReason
* messagesDeliveredConversationReason
* messagesDeliveredResendReason
* messagesDeliveredRegularReason
* messagesDeliveredInterpreterReason
* messagesMailed
* messagesSentViaHttp
* messagesFiltered
and 3 gauges for:
( Gauges are partial functions that run to grab a point in time statistic
when the stat is requested )
* users ( from Distributor )
* groups ""
* listeners ""
The stats MBean is dynamic and the stats are created when they are first
incremented, which means if you connect your jconsole at the start of the
server you may not see all the stats. You will need to reconnect later or
invoke the DynamicMBean.getBeanInfo method which jconsole only does once
when it connects.
There are two JMX operations: reset and getGatheringTime. Reset will reset
all statistics to zero and it will actually remove the gauges, which
shouldn't happen. It definitly needs some more work but I wanted to get some
opinions.
Here are a few items that can be added or improved.
* Timers
Timers are functions that wrap other functions, time how long it takes for
the wrapped function to execute and gathers max / min / avg.
Maybe wrap the message parser?.. Could also add other metrics like std
deviation and/or rate of increase/descrease in duration over time periods.
* Intigrate lift statistics
Lower level stats like http request counters and timers. Http response code
counters. DB connection pool counters and maybe a small array of the longest
running db queries.
* Add initializer stack for scala-stats
Add a mutable list of functions that will be called when the Stats object is
started or reset. This could initialize statistics from a db or elsewhere.
Let me know if this is the direction you guys are thinking of or if I am way
off mark.
Thanks
Andy
***************************
Index: src/main/scala/org/apache/esme/model/User.scala
===================================================================
--- src/main/scala/org/apache/esme/model/User.scala (revision 800477)
+++ src/main/scala/org/apache/esme/model/User.scala (working copy)
@@ -45,6 +45,8 @@
import java.net.URL
import java.util.logging._
+import com.twitter.service.{ Stats => ESMEStatistics }
+
object User extends User with KeyedMetaMapper[Long, User] {
override def afterSave = profileChanged _ :: notifyActors _ ::
super.afterSave
@@ -382,6 +384,9 @@
curUser.remove()
curUserId(Full(who.id.toString))
onLogIn.foreach(_(who))
+
+ // This may not be accurate because of users timing out instead of
logging out
+ ESMEStatistics incr "usersLoggedIn"
}
def logoutCurrentUser = logUserOut()
@@ -391,6 +396,7 @@
curUserId.remove()
curUser.remove()
S.request.foreach(_.request.getSession.invalidate)
+ ESMEStatistics getCounter "usersLoggedIn" decr
}
private object curUserId extends SessionVar[Box[String]](Empty)
Index: src/main/scala/org/apache/esme/lib/AccessPoolMgr.scala
===================================================================
--- src/main/scala/org/apache/esme/lib/AccessPoolMgr.scala (revision
800477)
+++ src/main/scala/org/apache/esme/lib/AccessPoolMgr.scala (working copy)
@@ -85,7 +85,7 @@
}
bind("add", in,
- "poolName" -> text("", addNewPool, "id" -> theInput)
+ "poolName" -> text("", addNewPool _, "id" -> theInput)
)
}
@@ -138,7 +138,7 @@
poolId.set(p.toLong);
redisplayPool()},
"id" -> editPoolName),
- "username" -> text(username, username = _, "id" -> editUsername),
+ "username" -> text(username, username = _:String, "id" ->
editUsername),
"permission" -> select(permissions, Empty, addPoolUser, "id" ->
editPermission)
)
Index: src/main/scala/org/apache/esme/lib/ActionMgr.scala
===================================================================
--- src/main/scala/org/apache/esme/lib/ActionMgr.scala (revision 800477)
+++ src/main/scala/org/apache/esme/lib/ActionMgr.scala (working copy)
@@ -123,7 +123,7 @@
bind("main", in,
- "name" -> text(name, name = _, "id" -> mainName),
+ "name" -> text(name, name = _:String, "id" -> mainName),
"test" -> textarea(test, test = _, "id" -> mainTest),
"action" -> textarea("", doSave, "id" -> mainAction)
)
Index: src/main/scala/org/apache/esme/lib/AuthMgr.scala
===================================================================
--- src/main/scala/org/apache/esme/lib/AuthMgr.scala (revision 800477)
+++ src/main/scala/org/apache/esme/lib/AuthMgr.scala (working copy)
@@ -100,7 +100,7 @@
}
bind("main", in,
- "token" -> text("", addAuthToken, "id" -> theInput)
+ "token" -> text("", addAuthToken _, "id" -> theInput)
)
}
Index: src/main/scala/org/apache/esme/lib/TrackMgr.scala
===================================================================
--- src/main/scala/org/apache/esme/lib/TrackMgr.scala (revision 800477)
+++ src/main/scala/org/apache/esme/lib/TrackMgr.scala (working copy)
@@ -101,7 +101,7 @@
}
bind("main", in,
- "track" -> text("", addTrack, "id" -> theInput)
+ "track" -> text("", addTrack _, "id" -> theInput)
)
}
Index: src/main/scala/org/apache/esme/actor/GroupActor.scala
===================================================================
--- src/main/scala/org/apache/esme/actor/GroupActor.scala (revision
800477)
+++ src/main/scala/org/apache/esme/actor/GroupActor.scala (working copy)
@@ -30,6 +30,7 @@
import org.apache.esme._
import model._
+import com.twitter.service.{ Stats => ESMEStatistics }
object GroupActor {
case class StartMeUp(group: Long)
@@ -42,6 +43,7 @@
react {
case StartMeUp =>
link(ActorWatcher)
+ ESMEStatistics incr "groupCount"
}
}
}
Index: src/main/scala/org/apache/esme/actor/ScalaInterpreter.scala
===================================================================
--- src/main/scala/org/apache/esme/actor/ScalaInterpreter.scala (revision
800477)
+++ src/main/scala/org/apache/esme/actor/ScalaInterpreter.scala (working
copy)
@@ -31,6 +31,8 @@
import org.apache.esme.model._
import net.liftweb.util.{Empty,Props}
+import com.twitter.service.{ Stats => ESMEStatistics }
+
object ScalaInterpreter extends Actor{
val settings = new Settings()
@@ -57,6 +59,9 @@
Distributor !
Distributor.AddMessageToMailbox(from, msg, InterpreterReason(from))
}
}
+
+ ESMEStatistics incr "interpreterMessagesCreated"
+ ESMEStatistics incr "messagesCreated"
}
}
Index: src/main/scala/org/apache/esme/actor/SchedulerActor.scala
===================================================================
--- src/main/scala/org/apache/esme/actor/SchedulerActor.scala (revision
800477)
+++ src/main/scala/org/apache/esme/actor/SchedulerActor.scala (working
copy)
@@ -28,6 +28,7 @@
import org.apache.esme.model._
import net.liftweb.http.ActorWatcher
import net.liftweb.util.{Full,Empty,TimeHelpers}
+import com.twitter.service.{ Stats => ESMEStatistics }
class SchedulerActor(val messageProcessor: Actor, val user: Long, val
everySeconds: Int, val reason: MailboxReason) extends Actor {
@@ -64,6 +65,8 @@
messageProcessor !
Distributor.AddMessageToMailbox(user, msg, reason)
// }
}
+ ESMEStatistics incr "schedulerMessagesCreated"
+ ESMEStatistics incr "messagesCreated"
}
}
Index: src/main/scala/org/apache/esme/actor/UserActor.scala
===================================================================
--- src/main/scala/org/apache/esme/actor/UserActor.scala (revision
800477)
+++ src/main/scala/org/apache/esme/actor/UserActor.scala (working copy)
@@ -36,6 +36,7 @@
import java.util.logging._
import java.util.{TimeZone, Calendar}
import scala.xml.{Elem}
+import com.twitter.service.{ Stats => ESMEStatistics }
object UserActor {
private[actor] case class StartMeUp(user: Long)
@@ -106,6 +107,8 @@
this ! UpdateTracking(Distributor.TrackTrackingType)
this ! UpdateTracking(Distributor.PerformTrackingType)
+ ESMEStatistics incr "userCount"
+
case RunFunc(f) =>
f()
@@ -135,7 +138,9 @@
else null
msg.saveMe
-
+ ESMEStatistics incr "userMessagesCreated"
+ ESMEStatistics incr "messagesCreated"
+
Distributor ! Distributor.AddMessageToMailbox(userId, msg,
NoReason)
for (id <- followers)
@@ -222,13 +227,20 @@
val mb = Mailbox.create.user(userId).message(msg)
reason match {
case TrackReason(trackId) => mb.viaTrack(trackId)
+ ESMEStatistics incr "messagesDeliveredTrackReason"
case DirectReason(fromId) => mb.directlyFrom(fromId)
+ ESMEStatistics incr "messagesDeliveredDirectReason"
case ConversationReason(convId) => mb.conversation(convId)
+ ESMEStatistics incr "messagesDeliveredConversationReason"
case ResendReason(resender) => mb.resentBy(resender)
+ ESMEStatistics incr "messagesDeliveredResendReason"
+ case RegularReason(id) => ESMEStatistics incr
"messagesDeliveredRegularReason"
+ case InterpreterReason(id) => ESMEStatistics incr
"messagesDeliveredInterpreterReason"
case _ =>
}
mb.saveMe
-
+ ESMEStatistics incr "messagesDelivered"
+
_mailbox = ((msg.id.is, reason) ::
_mailbox.toList).take(500).toArray
listeners.foreach(_ ! MessageReceived(msg, reason))
@@ -238,12 +250,14 @@
td.whatToDo match {
case m @ MailTo(_, _) =>
- User.find(userId).foreach( u =>
- HttpSender ! HttpSender.SendAMessage(m, msg, u, reason,
td.uniqueId))
+ User.find(userId).foreach( u => {
+ HttpSender ! HttpSender.SendAMessage(m, msg, u, reason,
td.uniqueId)
+ ESMEStatistics incr "messagesMailed"})
case h @ HttpTo(_, _, _, _, _) =>
- User.find(userId).foreach( u =>
- HttpSender ! HttpSender.SendAMessage(h, msg, u, reason,
td.uniqueId))
+ User.find(userId).foreach( u => {
+ HttpSender ! HttpSender.SendAMessage(h, msg, u, reason,
td.uniqueId)
+ ESMEStatistics incr "messagesSentViaHTTP"})
case PerformResend =>
if (! msg.saved_?) msg.save
@@ -256,7 +270,7 @@
case ScalaInterpret => if (msg.source.is != "scala")
ScalaInterpreter ! ScalaInterpreter.ScalaExcerpt(userId,
msg.id.is, msg.pool.is, msg.getText)
- case PerformFilter => // IGNORE
+ case PerformFilter => ESMEStatistics incr "messagesFiltered" //
IGNORE
}
}
}
Index: src/main/scala/org/apache/esme/actor/Distributor.scala
===================================================================
--- src/main/scala/org/apache/esme/actor/Distributor.scala (revision
800477)
+++ src/main/scala/org/apache/esme/actor/Distributor.scala (working copy)
@@ -159,4 +159,8 @@
ret
}
}
+
+ def getUsersCount = users.size
+ def getGroupsCount = groups.size
+ def getListenersCount = listeners.size
}
Index: src/main/scala/org/apache/esme/actor/MessagePullActor.scala
===================================================================
--- src/main/scala/org/apache/esme/actor/MessagePullActor.scala (revision
800477)
+++ src/main/scala/org/apache/esme/actor/MessagePullActor.scala (working
copy)
@@ -25,6 +25,7 @@
import scala.actors.Actor._
import net.liftweb.http.ActorWatcher
import org.apache.esme.actor.Distributor.{UserCreatedMessage=>Msg}
+import com.twitter.service.{ Stats => ESMEStatistics }
class MessagePullActor(val messageProcessor: Actor, private var
lastMessage: Option[Msg], val messageSource: UniqueMessageSource) extends
Actor {
@@ -45,6 +46,7 @@
for (message <- lastMessages) {
messageProcessor ! message
lastMessage = Some(message)
+ ESMEStatistics incr "messagesPulled"
}
}
case FetchMessages => actor {
Index: src/main/scala/bootstrap/liftweb/Boot.scala
===================================================================
--- src/main/scala/bootstrap/liftweb/Boot.scala (revision 800477)
+++ src/main/scala/bootstrap/liftweb/Boot.scala (working copy)
@@ -43,7 +43,7 @@
import org.compass.core.config.CompassConfiguration
import scala.actors.Actor
import Actor._
-
+import com.twitter.service.{StatsMBean, Stats => ESMEStatistics }
/**
* A class that's instantiated early and run. It allows the application
* to modify lift's environment
@@ -140,11 +140,18 @@
LiftRules.early.append(makeUtf8)
+ // register stats gathering object with platform mbean server
+ StatsMBean("org.apache.esme.stats")
+
Distributor.touch
SchedulerActor.touch
MessagePullActor.touch
ScalaInterpreter.touch
+ ESMEStatistics.makeGauge("users") { Distributor.getUsersCount }
+ ESMEStatistics.makeGauge("groups") { Distributor.getGroupsCount }
+ ESMEStatistics.makeGauge("listeners") { Distributor.getListenersCount }
+
Action.findAll(By(Action.disabled, false), By(Action.removed,
false)).foreach {
_.startActors
}
@@ -265,6 +272,7 @@
loop {
react {
case SessionWatcherInfo(sessions) =>
+ ESMEStatistics.getCounter("liftSessions").set(sessions.size)
if ((millis - tenMinutes) > lastTime) {
lastTime = millis
val rt = Runtime.getRuntime
Index: pom.xml
===================================================================
--- pom.xml (revision 800477)
+++ pom.xml (working copy)
@@ -137,6 +137,13 @@
<scope>provided</scope>
</dependency>
+ <!-- for stats gathering and stats MBean code can be found at
+ git://github.com/andythedestroyer/scala-stats.git -->
+ <dependency>
+ <groupId>com.twitter.service</groupId>
+ <artifactId>stats</artifactId>
+ <version>1.0</version>
+ </dependency>
<dependency>
<groupId>junit</groupId>
Re: Stats code
Posted by Andy the destroyer <an...@gmail.com>.
This is copied from Vassil's response another thread.
Vassil Dichev to esme-dev
show details 7:19 AM (5 hours ago)
Reply
[image: Follow up message]
> Latest code drop is deployed.
Ah, close miss for my latest minor fix :)
> Doesn't include andy's JMX code yet :-< Haven't quite figured out how
> to do this was an elaborate cut and paste action.
>From command-line:
patch -p0 < patch.diff
>From eclipse:
Right-click on parent directory, Team -> Apply patch
The latest commits might have accumulated conflicts with the patch,
but shouldn't be difficult to do by hand for one or two conflicts. The
problem with Andy's code currently is that there's no public maven
repository where we can find Andy's modified library, where the bulk
of the code is.
..
That should work .. I can also commit my branch to my github account. You
could check it out there without having to worry about patchfiles.
Let me know..
-Andy
On Mon, Aug 10, 2009 at 7:08 AM, Richard Hirsch <hi...@gmail.com>wrote:
> I'd like to try out Andy's code but haven't figured out how do it
> without editing my way to insanity.
>
> Does someone else have some sort of a tool (SVN-related??) that takes
> a svn diff file and edits the source files to contain the changes?
>
> D.
>
> On Fri, Aug 7, 2009 at 11:56 AM, Richard Hirsch<hi...@gmail.com>
> wrote:
> > Thanks. I'll try and create a deployment based on the patch and deploy
> > it to Stax on Monday.
> >
> > D.
> >
> > On Thu, Aug 6, 2009 at 11:31 PM, Andy the
> > destroyer<an...@gmail.com> wrote:
> >> I uploaded the patch and modified scala-stats.jar to JIRA (
> >> https://issues.apache.org/jira/browse/ESME-18).
> >>
> >> -Andy
> >>
> >> On Wed, Aug 5, 2009 at 12:11 PM, Andy the destroyer <
> >> andythedestroyer@gmail.com> wrote:
> >>
> >>> Hi Vassil,
> >>>
> >>> If you look at the scala-stats library, you need to look at my fork of
> it
> >>> on my git-hub account (http://github.com/andythedestroyer). I added
> some
> >>> more stats and JMX functionality that ESME is using. I provided a link
> note
> >>> in the pom section of the patch file. I am in contact with stevej it
> looks
> >>> like he is going to merge the changes I made into his master branch; I
> just
> >>> need to finish two more features and write the tests for them.
> >>>
> >>> I am also working on some actors for the scala-stats lib. I would be
> >>> interested to see your approach regarding stats and actors.
> >>>
> >>> I agree with you that retrieving stats with messages is a good idea. I
> even
> >>> started to look at what I would need to do to implement it. However,
> going
> >>> through the MsgParser was a nightmare. I am relatively new to Scala and
> very
> >>> new to parser combinators. With no comments in the code I found it very
> hard
> >>> to navigate my way through it and I moved on to other things.
> >>>
> >>> As far as licensing goes, scala-stats is under apache license and aside
> >>> from specs only has one dependecy-- configgy, which is also under
> Apache and
> >>> it's only dependecy is MarkdownJ which is BSD. Actually, the configgy
> dep is
> >>> optional. We don't even need it to run with ESME. Although, it might be
> >>> worth looking at configgy to manage ESME configs. Just a thought..
> >>>
> >>> I will submit the patch to the Jira ticket.
> >>>
> >>> Keep the faith,
> >>> Andy
> >>>
> >>>
> >>>
> >>> On Wed, Aug 5, 2009 at 4:42 AM, Vassil Dichev <vd...@apache.org>
> wrote:
> >>>
> >>>> Thanks Andy, this looks like an efficient use of an external library.
> >>>> I just heard of stevej's library because it's used in Jonas Boner's
> >>>> new Akka actors library. I need to take a look into scala-stats and
> >>>> see what else we can use from this library.
> >>>>
> >>>> Incidentally, I'm also writing an actor for popularity stats- e.g.
> >>>> which messages have been resent the most in the last time period, or
> >>>> which links have been clicked the most, or which users have been using
> >>>> a certain tag/word, etc. The statistics from this actor should be
> >>>> available in a UI snippet for the user to see, but I see no reason
> >>>> that it shouldn't also be available via JMX.
> >>>>
> >>>> As an aside, we could also make an action in a way that we can emulate
> >>>> a JMX console bot within ESME. For instance, you write a message with
> >>>> text "stats usersLoggedIn" and you get back a message with the user
> >>>> count.
> >>>>
> >>>> I guess after trying the patch we need to check the licenses of the
> >>>> used scala-stats library *and* its dependencies (if any). I am not
> >>>> sure what the exact procedure is for accepting a patch- I think we
> >>>> need a JIRA item with an attached patch where you click a license
> >>>> agreement.
> >>>>
> >>>> Thanks again, it's nice to see find ESME fun to work on!
> >>>>
> >>>> Vassil
> >>>>
> >>>
> >>>
> >>
> >
>
Re: Stats code
Posted by Richard Hirsch <hi...@gmail.com>.
I'd like to try out Andy's code but haven't figured out how do it
without editing my way to insanity.
Does someone else have some sort of a tool (SVN-related??) that takes
a svn diff file and edits the source files to contain the changes?
D.
On Fri, Aug 7, 2009 at 11:56 AM, Richard Hirsch<hi...@gmail.com> wrote:
> Thanks. I'll try and create a deployment based on the patch and deploy
> it to Stax on Monday.
>
> D.
>
> On Thu, Aug 6, 2009 at 11:31 PM, Andy the
> destroyer<an...@gmail.com> wrote:
>> I uploaded the patch and modified scala-stats.jar to JIRA (
>> https://issues.apache.org/jira/browse/ESME-18).
>>
>> -Andy
>>
>> On Wed, Aug 5, 2009 at 12:11 PM, Andy the destroyer <
>> andythedestroyer@gmail.com> wrote:
>>
>>> Hi Vassil,
>>>
>>> If you look at the scala-stats library, you need to look at my fork of it
>>> on my git-hub account (http://github.com/andythedestroyer). I added some
>>> more stats and JMX functionality that ESME is using. I provided a link note
>>> in the pom section of the patch file. I am in contact with stevej it looks
>>> like he is going to merge the changes I made into his master branch; I just
>>> need to finish two more features and write the tests for them.
>>>
>>> I am also working on some actors for the scala-stats lib. I would be
>>> interested to see your approach regarding stats and actors.
>>>
>>> I agree with you that retrieving stats with messages is a good idea. I even
>>> started to look at what I would need to do to implement it. However, going
>>> through the MsgParser was a nightmare. I am relatively new to Scala and very
>>> new to parser combinators. With no comments in the code I found it very hard
>>> to navigate my way through it and I moved on to other things.
>>>
>>> As far as licensing goes, scala-stats is under apache license and aside
>>> from specs only has one dependecy-- configgy, which is also under Apache and
>>> it's only dependecy is MarkdownJ which is BSD. Actually, the configgy dep is
>>> optional. We don't even need it to run with ESME. Although, it might be
>>> worth looking at configgy to manage ESME configs. Just a thought..
>>>
>>> I will submit the patch to the Jira ticket.
>>>
>>> Keep the faith,
>>> Andy
>>>
>>>
>>>
>>> On Wed, Aug 5, 2009 at 4:42 AM, Vassil Dichev <vd...@apache.org> wrote:
>>>
>>>> Thanks Andy, this looks like an efficient use of an external library.
>>>> I just heard of stevej's library because it's used in Jonas Boner's
>>>> new Akka actors library. I need to take a look into scala-stats and
>>>> see what else we can use from this library.
>>>>
>>>> Incidentally, I'm also writing an actor for popularity stats- e.g.
>>>> which messages have been resent the most in the last time period, or
>>>> which links have been clicked the most, or which users have been using
>>>> a certain tag/word, etc. The statistics from this actor should be
>>>> available in a UI snippet for the user to see, but I see no reason
>>>> that it shouldn't also be available via JMX.
>>>>
>>>> As an aside, we could also make an action in a way that we can emulate
>>>> a JMX console bot within ESME. For instance, you write a message with
>>>> text "stats usersLoggedIn" and you get back a message with the user
>>>> count.
>>>>
>>>> I guess after trying the patch we need to check the licenses of the
>>>> used scala-stats library *and* its dependencies (if any). I am not
>>>> sure what the exact procedure is for accepting a patch- I think we
>>>> need a JIRA item with an attached patch where you click a license
>>>> agreement.
>>>>
>>>> Thanks again, it's nice to see find ESME fun to work on!
>>>>
>>>> Vassil
>>>>
>>>
>>>
>>
>
Re: Stats code
Posted by Richard Hirsch <hi...@gmail.com>.
Thanks. I'll try and create a deployment based on the patch and deploy
it to Stax on Monday.
D.
On Thu, Aug 6, 2009 at 11:31 PM, Andy the
destroyer<an...@gmail.com> wrote:
> I uploaded the patch and modified scala-stats.jar to JIRA (
> https://issues.apache.org/jira/browse/ESME-18).
>
> -Andy
>
> On Wed, Aug 5, 2009 at 12:11 PM, Andy the destroyer <
> andythedestroyer@gmail.com> wrote:
>
>> Hi Vassil,
>>
>> If you look at the scala-stats library, you need to look at my fork of it
>> on my git-hub account (http://github.com/andythedestroyer). I added some
>> more stats and JMX functionality that ESME is using. I provided a link note
>> in the pom section of the patch file. I am in contact with stevej it looks
>> like he is going to merge the changes I made into his master branch; I just
>> need to finish two more features and write the tests for them.
>>
>> I am also working on some actors for the scala-stats lib. I would be
>> interested to see your approach regarding stats and actors.
>>
>> I agree with you that retrieving stats with messages is a good idea. I even
>> started to look at what I would need to do to implement it. However, going
>> through the MsgParser was a nightmare. I am relatively new to Scala and very
>> new to parser combinators. With no comments in the code I found it very hard
>> to navigate my way through it and I moved on to other things.
>>
>> As far as licensing goes, scala-stats is under apache license and aside
>> from specs only has one dependecy-- configgy, which is also under Apache and
>> it's only dependecy is MarkdownJ which is BSD. Actually, the configgy dep is
>> optional. We don't even need it to run with ESME. Although, it might be
>> worth looking at configgy to manage ESME configs. Just a thought..
>>
>> I will submit the patch to the Jira ticket.
>>
>> Keep the faith,
>> Andy
>>
>>
>>
>> On Wed, Aug 5, 2009 at 4:42 AM, Vassil Dichev <vd...@apache.org> wrote:
>>
>>> Thanks Andy, this looks like an efficient use of an external library.
>>> I just heard of stevej's library because it's used in Jonas Boner's
>>> new Akka actors library. I need to take a look into scala-stats and
>>> see what else we can use from this library.
>>>
>>> Incidentally, I'm also writing an actor for popularity stats- e.g.
>>> which messages have been resent the most in the last time period, or
>>> which links have been clicked the most, or which users have been using
>>> a certain tag/word, etc. The statistics from this actor should be
>>> available in a UI snippet for the user to see, but I see no reason
>>> that it shouldn't also be available via JMX.
>>>
>>> As an aside, we could also make an action in a way that we can emulate
>>> a JMX console bot within ESME. For instance, you write a message with
>>> text "stats usersLoggedIn" and you get back a message with the user
>>> count.
>>>
>>> I guess after trying the patch we need to check the licenses of the
>>> used scala-stats library *and* its dependencies (if any). I am not
>>> sure what the exact procedure is for accepting a patch- I think we
>>> need a JIRA item with an attached patch where you click a license
>>> agreement.
>>>
>>> Thanks again, it's nice to see find ESME fun to work on!
>>>
>>> Vassil
>>>
>>
>>
>
Re: Stats code
Posted by Andy the destroyer <an...@gmail.com>.
I uploaded the patch and modified scala-stats.jar to JIRA (
https://issues.apache.org/jira/browse/ESME-18).
-Andy
On Wed, Aug 5, 2009 at 12:11 PM, Andy the destroyer <
andythedestroyer@gmail.com> wrote:
> Hi Vassil,
>
> If you look at the scala-stats library, you need to look at my fork of it
> on my git-hub account (http://github.com/andythedestroyer). I added some
> more stats and JMX functionality that ESME is using. I provided a link note
> in the pom section of the patch file. I am in contact with stevej it looks
> like he is going to merge the changes I made into his master branch; I just
> need to finish two more features and write the tests for them.
>
> I am also working on some actors for the scala-stats lib. I would be
> interested to see your approach regarding stats and actors.
>
> I agree with you that retrieving stats with messages is a good idea. I even
> started to look at what I would need to do to implement it. However, going
> through the MsgParser was a nightmare. I am relatively new to Scala and very
> new to parser combinators. With no comments in the code I found it very hard
> to navigate my way through it and I moved on to other things.
>
> As far as licensing goes, scala-stats is under apache license and aside
> from specs only has one dependecy-- configgy, which is also under Apache and
> it's only dependecy is MarkdownJ which is BSD. Actually, the configgy dep is
> optional. We don't even need it to run with ESME. Although, it might be
> worth looking at configgy to manage ESME configs. Just a thought..
>
> I will submit the patch to the Jira ticket.
>
> Keep the faith,
> Andy
>
>
>
> On Wed, Aug 5, 2009 at 4:42 AM, Vassil Dichev <vd...@apache.org> wrote:
>
>> Thanks Andy, this looks like an efficient use of an external library.
>> I just heard of stevej's library because it's used in Jonas Boner's
>> new Akka actors library. I need to take a look into scala-stats and
>> see what else we can use from this library.
>>
>> Incidentally, I'm also writing an actor for popularity stats- e.g.
>> which messages have been resent the most in the last time period, or
>> which links have been clicked the most, or which users have been using
>> a certain tag/word, etc. The statistics from this actor should be
>> available in a UI snippet for the user to see, but I see no reason
>> that it shouldn't also be available via JMX.
>>
>> As an aside, we could also make an action in a way that we can emulate
>> a JMX console bot within ESME. For instance, you write a message with
>> text "stats usersLoggedIn" and you get back a message with the user
>> count.
>>
>> I guess after trying the patch we need to check the licenses of the
>> used scala-stats library *and* its dependencies (if any). I am not
>> sure what the exact procedure is for accepting a patch- I think we
>> need a JIRA item with an attached patch where you click a license
>> agreement.
>>
>> Thanks again, it's nice to see find ESME fun to work on!
>>
>> Vassil
>>
>
>
Re: Stats code
Posted by Andy the destroyer <an...@gmail.com>.
Hi Vassil,
If you look at the scala-stats library, you need to look at my fork of it on
my git-hub account (http://github.com/andythedestroyer). I added some more
stats and JMX functionality that ESME is using. I provided a link note in
the pom section of the patch file. I am in contact with stevej it looks like
he is going to merge the changes I made into his master branch; I just need
to finish two more features and write the tests for them.
I am also working on some actors for the scala-stats lib. I would be
interested to see your approach regarding stats and actors.
I agree with you that retrieving stats with messages is a good idea. I even
started to look at what I would need to do to implement it. However, going
through the MsgParser was a nightmare. I am relatively new to Scala and very
new to parser combinators. With no comments in the code I found it very hard
to navigate my way through it and I moved on to other things.
As far as licensing goes, scala-stats is under apache license and aside from
specs only has one dependecy-- configgy, which is also under Apache and it's
only dependecy is MarkdownJ which is BSD. Actually, the configgy dep is
optional. We don't even need it to run with ESME. Although, it might be
worth looking at configgy to manage ESME configs. Just a thought..
I will submit the patch to the Jira ticket.
Keep the faith,
Andy
On Wed, Aug 5, 2009 at 4:42 AM, Vassil Dichev <vd...@apache.org> wrote:
> Thanks Andy, this looks like an efficient use of an external library.
> I just heard of stevej's library because it's used in Jonas Boner's
> new Akka actors library. I need to take a look into scala-stats and
> see what else we can use from this library.
>
> Incidentally, I'm also writing an actor for popularity stats- e.g.
> which messages have been resent the most in the last time period, or
> which links have been clicked the most, or which users have been using
> a certain tag/word, etc. The statistics from this actor should be
> available in a UI snippet for the user to see, but I see no reason
> that it shouldn't also be available via JMX.
>
> As an aside, we could also make an action in a way that we can emulate
> a JMX console bot within ESME. For instance, you write a message with
> text "stats usersLoggedIn" and you get back a message with the user
> count.
>
> I guess after trying the patch we need to check the licenses of the
> used scala-stats library *and* its dependencies (if any). I am not
> sure what the exact procedure is for accepting a patch- I think we
> need a JIRA item with an attached patch where you click a license
> agreement.
>
> Thanks again, it's nice to see find ESME fun to work on!
>
> Vassil
>
Re: Stats code
Posted by Vassil Dichev <vd...@apache.org>.
Thanks Andy, this looks like an efficient use of an external library.
I just heard of stevej's library because it's used in Jonas Boner's
new Akka actors library. I need to take a look into scala-stats and
see what else we can use from this library.
Incidentally, I'm also writing an actor for popularity stats- e.g.
which messages have been resent the most in the last time period, or
which links have been clicked the most, or which users have been using
a certain tag/word, etc. The statistics from this actor should be
available in a UI snippet for the user to see, but I see no reason
that it shouldn't also be available via JMX.
As an aside, we could also make an action in a way that we can emulate
a JMX console bot within ESME. For instance, you write a message with
text "stats usersLoggedIn" and you get back a message with the user
count.
I guess after trying the patch we need to check the licenses of the
used scala-stats library *and* its dependencies (if any). I am not
sure what the exact procedure is for accepting a patch- I think we
need a JIRA item with an attached patch where you click a license
agreement.
Thanks again, it's nice to see find ESME fun to work on!
Vassil