java.lang.NumberFormatException:For input string: ""

[#1]

Error

A server error has occurred. Send the following information to Cloudera.

Path: http://slnxcloudera01.duns.uk.sopra:7180/cmf/parcel/details

Version: Cloudera Express 5.2.1 (#109 built by jenkins on 20141125-0809 git: 8b79f749fed7e9cef6c3c9cb0664ac1c28b7f9a4)

java.lang.NumberFormatException:For input string: ""
at NumberFormatException.java line 65
in java.lang.NumberFormatException forInputString()

Stack Trace:

1. NumberFormatException.java line 65
   in java.lang.NumberFormatException forInputString()
2. Long.java line 453
   in java.lang.Long parseLong()
3. Long.java line 540
   in java.lang.Long valueOf()
4. ParcelHelpers.java line 363
   in com.cloudera.parcel.ParcelHelpers getMajorVersion()
5. ParcelClusterUpgradeStatus.java line 43
   in com.cloudera.parcel.ParcelClusterUpgradeStatus evaluateParcel()
6. ParcelClusterUpgradeStatus.java line 76
   in com.cloudera.parcel.ParcelClusterUpgradeStatus of()
7. ClusterParcelStatus.java line 407
   in com.cloudera.parcel.ClusterParcelStatus of()

I get this error repeatedly when using the UI.

I believe it has been raised before with no resolution:

http://grokbase.com/t/cloudera/scm-users/13cg5rhc2h/cloudera-server-error

It doesn't matter if I install with parcels or packages the result is the same.

Thanks

[#2]

Are you able to provide the following listing from the CM host please:
# ls -l /opt/cloudera/parcel*

[#3]

See below.
The error appears during the wizard before the parcels have been selected. Although previously when I installed using the package approach I got as far as completing the wizard and then the error appeared when logging into CM for the first time.
The build is CM 5.2.1 on CentOS 6.3.
Thanks

/opt/cloudera/parcel-cache:
total 0

/opt/cloudera/parcel-repo:
total 0

/opt/cloudera/parcels:
total 4
drwxr-xr-x. 3 root root 4096 Jul 7 16:13 CDH

[#4 · Accepted solution]

Here is the listing from my host. This is a worker node but should look similar on the Cloudera Manager host as well

# ls -l /opt/cloudera/parcel*
/opt/cloudera/parcel-cache:
total 2990852
-rw-r--r-- 1 root root 1531811730 Dec 10 01:10 CDH-5.2.1-1.cdh5.2.1.p0.12-el6.parcel
-rw-r--r-- 1 root root 1530807851 Dec 10 00:46 CDH-5.2.1-1.cdh5.2.1.p507.148-el6.parcel

/opt/cloudera/parcel-repo:
total 2990860
-rw-r----- 1 cloudera-scm cloudera-scm 1531811730 Dec 10 00:37 CDH-5.2.1-1.cdh5.2.1.p0.12-el6.parcel
-rw-r----- 1 cloudera-scm cloudera-scm         42 Dec 10 00:37 CDH-5.2.1-1.cdh5.2.1.p0.12-el6.parcel.sha
-rw-r----- 1 cloudera-scm cloudera-scm 1530807851 Dec 10 00:45 CDH-5.2.1-1.cdh5.2.1.p507.148-el6.parcel
-rw-r----- 1 cloudera-scm cloudera-scm         41 Dec 10 00:45 CDH-5.2.1-1.cdh5.2.1.p507.148-el6.parcel.sha

/opt/cloudera/parcels:
total 12
-rw-r--r--  1 root root 2336 Dec 10 01:40 a
lrwxrwxrwx  1 root root   29 Dec 10 01:32 CDH -> CDH-5.2.1-1.cdh5.2.1.p507.148
drwxrwxr-x 11 root root 4096 Nov 19 23:18 CDH-5.2.1-1.cdh5.2.1.p0.12
drwxrwxr-x 11 root root 4096 Dec  3 01:25 CDH-5.2.1-1.cdh5.2.1.p507.148

The exception you're seeing is due to the parse of a parcel name failing. The format should be like in the listing above whereas in your cluster it is just "CDH" which should ideally be a symlink, not a directory.

Are you aware if the contents in these directories was manually changed sometime?

[#5]

That CM folder contained by the looks of things an old Spark parcel from a previous failed install.
Removing the CM folder (in fact the /opt/cloudera/ folder in its entirety) and removing and reinstalling cloudera-scm-manager seems to have resolved the issue.
It would be nice to see this error handled better so it doesn't bring back the endless pop ups. And maybe the documentation updated if that folder does need to be empty prior to a new installation.
Thanks

[#6]

Your point is taken. I will file an internal jira to track this error as it is confusing. The error message itself could use some improvement.