WMS 3.4.1 EMI2 precertification report

Testing server on: cream-51.pd.infn.it (SL5_64)

First installed and configured a EMI2 UI. Then installed update to glite-wms-ui-command (glite-wms-ui-commands-3.4.0-5.sl5.x86_64.rpm) from this repository:

http://etics-repository.cern.ch/repository/pm/registered/repomd/id/685c7797-b6c2-4587-964e-b3d479998aa2/sl5_x86_64_gcc412EPEL/index.html

built from WMS UI etics configuration emi-wms-ui_R_3_4_0_6.

Tried some simple job submission (without and with ISB)

dorigoa@cream-51 17:12:35 ~>rpm -qa | grep commands|grep wms
glite-wms-ui-commands-3.4.0-5.sl5

dorigoa@cream-51 17:12:41 ~>glite-wms-job-submit -a -e https://cream-01.pd.infn.it:7443/glite_wms_wmproxy_server sleep.jdl

Connecting to the service https://cream-01.pd.infn.it:7443/glite_wms_wmproxy_server


====================== glite-wms-job-submit Success ======================

The job has been successfully submitted to the WMProxy
Your job identifier is:

https://devel09.cnaf.infn.it:9000/VfRdyI099ZXzjrtGfE3QYA

==========================================================================


dorigoa@cream-51 17:12:43 ~>glite-wms-job-status https://devel09.cnaf.infn.it:9000/VfRdyI099ZXzjrtGfE3QYA


======================= glite-wms-job-status Success =====================

BOOKKEEPING INFORMATION:

Status info for the Job : https://devel09.cnaf.infn.it:9000/VfRdyI099ZXzjrtGfE3QYA
Current Status:     Ready
Destination:        lcgceatlas.dnp.fmph.uniba.sk:8443/cream-pbs-dteam
Submitted:          Mon Jan 21 17:12:43 2013 CET
==========================================================================

dorigoa@cream-51 17:12:51 ~>cat sleep.jdl 
[
Executable = "/bin/echo";
Arguments = "ciao";
MyProxyServer="";
]

dorigoa@cream-51 17:14:06 ~>cat wms_ISB.jdl 
[
Executable = "/bin/sleep" ;
Arguments = "15";
#InputSandbox =  {"/etc/group"};
InputSandbox = {"/home/dorigoa/mysandbox.txt", "/home/dorigoa/mysandbox2.txt"};
#InputSandboxBaseURI = "gsiftp://cream-23.pd.infn.it/var/";
AllowZippedISB = true;
requirements = RegExp("cream.*", other.GlueCEUniqueID);
RetryCount = 0;
ShallowRetryCount = 1;
]

dorigoa@cream-51 17:14:25 ~>glite-wms-job-submit -a -e https://cream-01.pd.infn.it:7443/glite_wms_wmproxy_server wms_ISB.jdl 

Connecting to the service https://cream-01.pd.infn.it:7443/glite_wms_wmproxy_server


====================== glite-wms-job-submit Success ======================

The job has been successfully submitted to the WMProxy
Your job identifier is:

https://devel09.cnaf.infn.it:9000/BnYdX-_NytVUbMopFuG18g

==========================================================================


dorigoa@cream-51 17:14:37 ~>glite-wms-job-status https://devel09.cnaf.infn.it:9000/BnYdX-_NytVUbMopFuG18g


======================= glite-wms-job-status Success =====================

BOOKKEEPING INFORMATION:

Status info for the Job : https://devel09.cnaf.infn.it:9000/BnYdX-_NytVUbMopFuG18g
Current Status:     Ready
Destination:        ce203.cern.ch:8443/cream-lsf-grid_2nh_dteam
Submitted:          Mon Jan 21 17:14:36 2013 CET
==========================================================================

LIST OF BUGS

UI

EMI-2 WMS UI cannot handle jobids starting with a '-'. (https://savannah.cern.ch/bugs/?98113)

BUGS:

EMI-2 WMS UI cannot handle jobids starting with a '-'. (https://savannah.cern.ch/bugs/?98113) PRE-CERTIFIED (adorigo 08/01/2013)

The bug has been fixed and verified. In order to check it, please take into account that there's a 2% probability to get the "-" as leading character of the gridjobid. By submitting (in a loop) about 1000 jobs, there's a 100% to get the at least 1 candidate job. The JDL to use is:

[
Type       = "Job";
JobType    = "Normal";
Executable = "/bin/hostname";
StdOutput  = "hello.out";
StdError   = "hello.err";
InputSandbox =  {"test_script_120.sh"};
OutputSandbox =  {"hello.out","hello.err"};
RetryCount = 0;
ShallowRetryCount  =  0;
Requirements =
  other.GlueSiteUniqueID == "CERN-PROD" ||
  other.GlueCEInfoHostName == "x";
]

Submit, in a loop, about 200-400 jobs with the above JDL. Redirect the output to a logfile; when the candidate GridJobID is generated the loop should stop and in the log file you can see a message like this:

terminate called after throwing an instance of 'boost::filesystem::filesystem_error'
  what():  boost::filesystem::path: invalid name "-V" in path: "SandboxDir/-V/https_3a_2f_2fwms004.cnaf.infn.it_3a9000_2f-VZRnkQj2MopgqpsvdLjRA/input/group"
./test_98113.sh: line 6: 19300 Aborted                 glite-wms-job-submit -o GIDS -d pippo -e https://cream-28.pd.infn.it:7443/glite_wms_wmproxy_server my.jdl

This is a loop that should ensure to get the candidate GridJobID:

#!/bin/bash
set -e
set i=0
for i in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
  do
    for k in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
      do
        glite-wms-job-submit -o GIDS -a -e https://<YOUR_PREFERRED_WMS_ENDPOINT>:7443/glite_wms_wmproxy_server <YOUR_JDL>.jdl
        sleep 1
      done
  done

It is better to pre-delegate a proxy and use the delegation ID with "-d" option of glite-wms-job-submit instead of "-a" (automatica delegation mean a delegation per job, that would take a lot of time).

After installing new rpm of glite-wms-ui-commands (version 3_4_0_5) the abort shown above shoud disappear.

-- AlviseDorigo - 2013-01-09