PfpGrowth is also ok with the retails an daccidents dataset(
http://fimi.ua.ac.be/data/) when runing over my cluster. My cluster has
1500+ nodes.
As my cluster is built with hadoop-0.19, I modified the mahout-0.5 PfpGrowth
MR job souce code to adapt with hadoop-0.19 by using the old hadoop
api(org.apache.hadoop.mapred.*), but i didn't change the Pfpgrowth algorithm
itself.
I don't think cache misses relate to distributed cache(if I am wrong, please
correct me), it relates to FPTreeDepthCache.
The status of the tasks which got stuck always showed* "Processing FPTree:
Bottom Up FP Growth reduce"*,
so I logged the status within FpGrowth.growthBottomUp().
* updater.update("Bottom Up FP Growth");
log.info("Bottom Up FP Growth");*
And then from logs, I found those stuck tasks was keep outputing *"Bottom Up
FP Growth".
*See *:
*
-09-22 21:58:35,689 INFO
org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Bottom Up FP Growth
2011-09-22 21:58:35,690 INFO
org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Bottom Up FP Growth
2011-09-22 21:58:35,690 INFO
org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Bottom Up FP Growth
2011-09-22 21:58:35,690 INFO
org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Bottom Up FP Growth
2011-09-22 21:58:35,690 INFO
org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Bottom Up FP Growth
2011-09-22 21:58:35,691 INFO
org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Bottom Up FP Growth
2011-09-22 21:58:35,692 INFO
org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Bottom Up FP Growth
2011-09-22 21:58:35,693 INFO
org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Bottom Up FP Growth
2011-09-22 21:58:35,694 INFO
org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Bottom Up FP Growth
2011-09-22 21:58:35,694 INFO
org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Bottom Up FP Growth
It's mostly likely FpGrowth.growthBottomUp() run into very big deep
recursion, or even possiblely endless loop.
2011/9/22 戴清灏 <[email protected]>
> Hi,
> The Pfp-growth on my cluster is fine when I run it onto
> *webdocs.dat*dataset.(
> http://fimi.ua.ac.be/data/)
> My cluster has about 15 nodes and this will only take less than 10 mins.
> I looked into you log and found that cache misses.
> Which version of mahout are you using?
> For mahout-0.4, it would not use distributed-cache provided by hadoop.
> For mahout-0.5, the code is a little bit different and it uses
> distributed cache, which will fetch FList and gList file before execution.
> (setup method, in reducer)
> For more info about distributed cache in hadoop, see
>
> http://hadoop.apache.org/common/docs/r0.20.2/api/org/apache/hadoop/filecache/DistributedCache.html
>
> Best,
> Roger Dai
>
> 2011/9/22 bing wang <[email protected]>
>
> > Hi all,
> >
> > I run PfpGrowth in a big hadoop cluster , but the ParallelFPgrowth MR job
> > got stuck when some ParallelFPGrowthReducer tasks run into
> > fpGrowth.generateTopKFrequentPatterns().
> >
> > I feed 6,468,284 record for PfpGrowth, and 110825 features greater than
> > minSupport(5), numGroups(20000), treeCacheSize(10) and with other
> paramters
> > as default.
> > I also set the hadoop mapred.child.java.opts option to 3g.
> >
> > The cluster started 20 ParallelFPGrowthReducer tasks, but about 10 tasks
> > was
> > stuck at the same reason.
> >
> > See the reducer task log(Log 1):
> >
> > 2011-09-20 21:11:49,860 INFO
> > org.apache.mahout.fpm.pfpgrowth.ParallelCountingReducer: localFList size
> > 871
> > for key 8120
> > 2011-09-20 21:11:49,860 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Number of unique items
> > 871
> > 2011-09-20 21:11:49,860 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> 48722=>3
> > 2011-09-20 21:11:49,860 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> 48723=>4
> > 2011-09-20 21:11:49,860 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> 48720=>5
> > 2011-09-20 21:11:49,860 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> 48721=>0
> > 2011-09-20 21:11:49,860 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> 48724=>2
> > 2011-09-20 21:11:49,860 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> 48725=>1
> > 2011-09-20 21:11:49,860 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Number of unique
> pruned
> > items 27
> > 2011-09-20 21:11:49,861 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Number of Nodes in the
> > FP
> > Tree: 93
> > 2011-09-20 21:11:49,861 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Mining FTree Tree for
> > all
> > patterns with 5
> > 2011-09-20 21:11:49,861 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Found 2 Patterns with
> > Least Support 10
> > 2011-09-20 21:11:49,861 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Mining FTree Tree for
> > all
> > patterns with 4
> > 2011-09-20 21:11:49,861 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Found 1 Patterns with
> > Least Support 13
> > 2011-09-20 21:11:49,861 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Mining FTree Tree for
> > all
> > patterns with 3
> > 2011-09-20 21:11:49,862 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Found 2 Patterns with
> > Least Support 13
> > 2011-09-20 21:11:49,862 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Mining FTree Tree for
> > all
> > patterns with 2
> > 2011-09-20 21:11:49,862 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Found 1 Patterns with
> > Least Support 14
> > 2011-09-20 21:11:49,862 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Mining FTree Tree for
> > all
> > patterns with 1
> > 2011-09-20 21:11:49,863 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Found 1 Patterns with
> > Least Support 15
> > 2011-09-20 21:11:49,863 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Mining FTree Tree for
> > all
> > patterns with 0
> > 2011-09-20 21:11:49,863 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Found 2 Patterns with
> > Least Support 17
> > 2011-09-20 21:11:49,863 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Tree Cache: First
> Level:
> > Cache hits=20 Cache Misses=127
> > 2011-09-20 21:11:49,864 INFO
> > org.apache.mahout.fpm.pfpgrowth.ParallelCountingReducer: localFList size
> > 1410 for key 8140
> > 2011-09-20 21:11:49,865 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Number of unique items
> > 1410
> > 2011-09-20 21:11:49,865 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> 48842=>22
> > 2011-09-20 21:11:49,865 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> 48843=>23
> > 2011-09-20 21:11:49,865 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> 48840=>21
> > 2011-09-20 21:11:49,865 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> 48841=>27
> > 2011-09-20 21:11:49,865 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> > 48844=>266
> > 2011-09-20 21:11:49,865 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> 48845=>28
> > 2011-09-20 21:11:49,865 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Number of unique
> pruned
> > items 353
> > 2011-09-20 21:11:49,866 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Number of Nodes in the
> > FP
> > Tree: 2512
> > 2011-09-20 21:11:49,866 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Mining FTree Tree for
> > all
> > patterns with 266
> >
> >
> > From last line of Log 1, we can see the reducer task which get stuck is
> > mining FTree Tree for all patterns with the attribute 266.
> >
> > However I think the task is still running because some reducer tasks will
> > keep moving after long time mining(see bellow log, Log 2)
> >
> > 2011-09-20 21:24:08,577 INFO
> > org.apache.mahout.fpm.pfpgrowth.ParallelCountingReducer: localFList size
> > 1183 for key 13596
> > 2011-09-20 21:24:08,578 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Number of unique items
> > 1183
> > 2011-09-20 21:24:08,578 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> > 81581=>250
> > 2011-09-20 21:24:08,578 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> > 81580=>132
> > 2011-09-20 21:24:08,578 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> > 81577=>129
> > 2011-09-20 21:24:08,578 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> > 81576=>128
> > 2011-09-20 21:24:08,578 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> > 81579=>131
> > 2011-09-20 21:24:08,578 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Adding Pattern
> > 81578=>130
> > 2011-09-20 21:24:08,578 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Number of unique
> pruned
> > items 251
> > 2011-09-20 21:24:08,579 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Number of Nodes in the
> > FP
> > Tree: 1376
> > 2011-09-20 21:24:08,579 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Mining FTree Tree for
> > all
> > patterns with 250
> > 2011-09-20 21:24:08,579 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Found 2 Patterns with
> > Least Support 5
> > 2011-09-20 21:24:08,579 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Mining FTree Tree for
> > all
> > patterns with 132 (take 2.5 hours )
> > 2011-09-21 00:08:56,616 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Found 40 Patterns with
> > Least Support 7
> > 2011-09-21 00:08:56,665 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Mining FTree Tree for
> > all
> > patterns with 131
> > 2011-09-21 00:08:56,703 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Found 3 Patterns with
> > Least Support 7
> > 2011-09-21 00:08:56,703 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Mining FTree Tree for
> > all
> > patterns with 130
> > 2011-09-21 00:08:56,736 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Found 2 Patterns with
> > Least Support 7
> > 2011-09-21 00:08:56,736 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Mining FTree Tree for
> > all
> > patterns with 129
> > 2011-09-21 00:08:56,769 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Found 1 Patterns with
> > Least Support 7
> > 2011-09-21 00:08:56,769 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Mining FTree Tree for
> > all
> > patterns with 128
> > 2011-09-21 00:08:56,809 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Found 1 Patterns with
> > Least Support 7
> > 2011-09-21 00:08:56,809 INFO
> > org.apache.mahout.fpm.pfpgrowth.fpgrowth.FPGrowth: Tree Cache: First
> Level:
> > Cache hits=13 Cache Misses=475
> >
> > And, the task status showed *"Processing FPTree: Bottom Up FP Growth >
> > reduce"*, which indicates that the job run into FpGrowth.growthBottomUp()
> > and be stuck within FpGrowth.growthBottomUp().
> >
> > Some reduers will hold at this status about 25 hours till I kill the
> whole
> > PfpGrowth job manually.(see job history as follows)
> >
> > *JobName: * PFP Parallel FPGrowth running over input
> (ignore)/sortedoutput
> > *JobConf: *
> >
> >
> hdfs://hdpnn:9000/home/hadoop/cluster-data/mapred/system/job_201108301259_1089136/job.xml<
> >
> http://10.249.54.103/jobconf_history.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&jobLogDir=file:/home/hadoop/cluster-data/logs/history&jobUniqueString=hdpjt_1314680340456_job_201108301259_1089136
> > >
> > *Submitted At: * 20-Sep-2011 20:31:57
> > *Launched At: * 20-Sep-2011 20:32:04 (6sec)
> > *Finished At: * 21-Sep-2011 22:31:51 (25hrs, 59mins, 47sec)
> > *Status: * KILLED
> > *Analyse This Job<
> >
> http://10.249.54.103/analysejobhistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp
> > >
> > *
> > ------------------------------
> >
> > Counter Map Reduce Total KindTotal
> > Tasks(successful+failed+killed)Successful
> > tasksFailed tasksKilled tasksStart TimeFinish Time Setup
> > 1<
> >
> http://10.249.54.103/jobtaskshistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&taskType=SETUP&status=all
> > >
> > 1<
> >
> http://10.249.54.103/jobtaskshistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&taskType=SETUP&status=SUCCESS
> > >
> > 0<
> >
> http://10.249.54.103/jobtaskshistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&taskType=SETUP&status=FAILED
> > >
> > 0<
> >
> http://10.249.54.103/jobtaskshistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&taskType=SETUP&status=KILLED
> > >
> > 20-Sep-2011
> > 20:32:06 20-Sep-2011 20:32:07 (0sec) Map
> > 28<
> >
> http://10.249.54.103/jobtaskshistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&taskType=MAP&status=all
> > >
> > 22<
> >
> http://10.249.54.103/jobtaskshistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&taskType=MAP&status=SUCCESS
> > >
> > 0<
> >
> http://10.249.54.103/jobtaskshistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&taskType=MAP&status=FAILED
> > >
> > 6<
> >
> http://10.249.54.103/jobtaskshistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&taskType=MAP&status=KILLED
> > >
> > 20-Sep-2011
> > 20:32:13 20-Sep-2011 20:56:01 (23mins, 48sec) Reduce
> > 24<
> >
> http://10.249.54.103/jobtaskshistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&taskType=REDUCE&status=all
> > >
> > 18<
> >
> http://10.249.54.103/jobtaskshistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&taskType=REDUCE&status=SUCCESS
> > >
> > 0<
> >
> http://10.249.54.103/jobtaskshistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&taskType=REDUCE&status=FAILED
> > >
> > 6<
> >
> http://10.249.54.103/jobtaskshistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&taskType=REDUCE&status=KILLED
> > >
> > 20-Sep-2011
> > 20:32:33 21-Sep-2011 22:31:51 (25hrs, 59mins, 18sec) Cleanup
> > 1<
> >
> http://10.249.54.103/jobtaskshistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&taskType=CLEANUP&status=all
> > >
> > 1<
> >
> http://10.249.54.103/jobtaskshistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&taskType=CLEANUP&status=SUCCESS
> > >
> > 0<
> >
> http://10.249.54.103/jobtaskshistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&taskType=CLEANUP&status=FAILED
> > >
> > 0<
> >
> http://10.249.54.103/jobtaskshistory.jsp?jobid=job_201108301259_1089136&logFile=file:/home/hadoop/cluster-data/logs/history/hdpjt_1314680340456_job_201108301259_1089136_lingfeng.wb_PFP%2BParallel%2BFPGrowth%2Brunning%2Bover%2Binput%252Fgroup%252Ftbp&taskType=CLEANUP&status=KILLED
> > >
> > 21-Sep-2011
> > 22:31:50 21-Sep-2011 22:31:51 (0sec)
> >
> >
> > In Log 2, we see FP Tree for key 13596 only includes 1376 nodes,
> > however it take 2.5 hours to ming patterns for 132.
> >
> > So why it was so slowly when mining frequent patterns for some
> attributes?
> > What's the reason behinds that or how can I resolve?
> >
> > Thanks in advance!
> >
> >
> >
> > --
> > Gmail/talk: [email protected]
> >
>
--
Gmail/talk: [email protected]
