The original version indeed only runs for two iterations, can you check your logs to see if anything wrong happens?

from wormhole.

ok, yes. i made a mistake.
thank you

from wormhole.

hi Tianqi,
Is there a way to check job progress like a web UI ?
Or else I have no idea what's going on.
I see that you provide both reg:linear and reg:logistic, For ranking
problem, which would you recommend?
Thanks a lot !

2015-11-26 14:51 GMT+08:00 Tianqi Chen [email protected]:

The original version indeed only runs for two iterations, can you check
your logs to see if anything wrong happens?

—
Reply to this email directly or view it on GitHub
#48 (comment).

from wormhole.

You can get into each of the workers, and check the stdout stderr logs of the worker

from wormhole.

from wormhole.

data=hdfs:///libsvmTrain/ should work, except that you need to make sure it is absolute path to the folder and the folder do not contain additional files

from wormhole.

I tried . An IOException is thrown.
1
2
3 Container: container_1448593333704_15438_01_000003 on
hadoop149.photo.163.org_54444

4

5 LogType: stderr
6 LogLength: 719
7 Log Contents:
8 connect to ip: [10.160.128.51]
9 start hadoop149.photo.163.org:19
10 readDirect: FSDataInputStream#read error:
11 java.io.IOException: Filesystem closed
12 at org.apache.hadoop.hdfs.DFSClient.checkOpen(DFSClient.java:730)
13 at
org.apache.hadoop.hdfs.DFSInputStream.readWithStrategy(DFSInputStream.java:779)
14 at
org.apache.hadoop.hdfs.DFSInputStream.read(DFSInputStream.java:847)
15 at
org.apache.hadoop.fs.FSDataInputStream.read(FSDataInputStream.java:143)
16 [13:46:35] include/dmlc/logging.h:208: [13:46:35]
src/io/hdfs_filesys.cc:44: HDFSStream.hdfsRead Error:Unknown error 255
17 terminate called after throwing an instance of 'dmlc::Error'
18 what(): [13:46:35] src/io/hdfs_filesys.cc:44: HDFSStream.hdfsRead
Error:Unknown error 255
19
20 LogType: stdout
21 LogLength: 0
22 Log Contents:

2015-11-28 2:09 GMT+08:00 Tianqi Chen [email protected]:

data=hdfs:///libsvmTrain/ should work, except that you need to make sure
it is absolute path to the folder and the folder do not contain additional
files

—
Reply to this email directly or view it on GitHub
#48 (comment).

from wormhole.

Just got the same error with 'Filesystem is closed'. It was because i manually killed application while it was working normal. I don't know, may be i iterrupted reading process and i got such message.
So, if somebody gets this error, just restart your job and wait longer, than the last time.

from wormhole.