[gridengine users] cygwin support by scalable logic

Raghavendra raghavendra at vizexperts.com
Mon Jul 1 10:55:41 UTC 2013


Hi Heraild

Small difference from earlier mail, I found that in windows we have to 
change permission on the port6444 folder manually. So after doing that 
the error message has changed from ssl certificate error (please check 
certificate validity) to ssl connect handshake timeout(connect timeout 
error).

waiting for your reply..


Ryagz

On 06/29/2013 01:30 PM, Raghavendra wrote:
> Hi Herald,
>
> Still no luck. There isnt any difference in doing with source or 
> executable way. I have gone through the inst_sge script, it trying to 
> get QMaster and EXECD port numbers by knowing their services. 
> Especially in settings.sh script i am just getting unset QMASTER_PORT 
> and EXECD_PORT no commands available to set or export them.
>
> One gud thing is, i have re installed every thing from scratch again 
> including SUA. Luckily the SSL library loading problem is resolved.  
> But when i give qconf -sconf i am getting a new error like :
>
> sh-3.1$ qconf -sconf
> error: commlib error: ssl certificate error (please check certificate 
> validity)
> ERROR: unable to contact qmaster using port 6444 on host "raghavendra"
>
> This error looks like there is a mismatch between the certificates. 
> But i have done exactly as said in the document. For the sake of 
> clarity i just want to describe wat i have done.
>
> 1. Installed SUA in windows, DEP Disabled.
> 2. set up debian-interix for SUA following the instructions in - 
> http://debian-interix.net/debian-interix/INSTALL
> 3. Tried installing OpenSSH Server using apt-get
>
>      #apt-get install openssh-server
>
>     got following error:
>
>     Unpacking openssh-blacklist (from 
> .../openssh-blacklist_0.4.1+nmu1_all.deb) ...
>     dpkg-deb: file 
> `/var/cache/apt/archives/openssh-blacklist_0.4.1+nmu1_all.deb' co
>     ntains ununderstood data member data.tar.xz     , giving up
>     dpkg: error processing 
> /var/cache/apt/archives/openssh-blacklist_0.4.1+nmu1_all.
>     deb (--unpack):
>      subprocess dpkg-deb --fsys-tarfile returned error exit status 2
>     Selecting previously deselected package openssh-server.
>     Unpacking openssh-server (from 
> .../openssh-server_1%3a4.7p1-9+interix.3_interix-
>     i386.deb) ...
>     Errors were encountered while processing:
> /var/cache/apt/archives/openssh-blacklist_0.4.1+nmu1_all.deb
>     E: Sub-process /usr/bin/dpkg returned an error code (1)
>
>     Now I am able to connect to other systems with SSH but other 
> systems cannot connect to mine.
>
> 4. Setup QMaster in a RHEL 64 bit linux system using the respective 
> common.tar and lx-amd64.tar files
>     Installed with CSP and did all that is required for windows 
> execution host. The certificate files were
>     created in /var/sgeCA/default/sge_QMaster folder but according to 
> document they should be created in p6444 folder
> 5. After Setting up QMaster in Linux system, I extracted common.tar 
> and windows specific.tar file in windows system. I copied the 
> installation folder "default" folder
>    to the SGE_Root directory of windows system.
> 6. I copied and extracted the certificate files generated in Linux 
> system to /var of Windows SUA.
> 7. In SUA shell sourced settings.sh present in default folder.
> 8. Run the qconf -sh or qconf -scong in SUA shell. Initailly it gave 
> me error that cannot find cert.pem file in 
> /var/sgeCA/port6444/default/userkeys/vizexperts directory.
> 9. My directory structure was different with sge_qmaster folder 
> instead of port6444 - /var/sgeCA/sge_qmaster/default/userkeys/vizexperts
> 10. so changed directory name of sge_qmaster to p6444 and ran qconf 
> -sconf again then i got the above mentioned SSL certificate error.
>
> I am not sure if i am doing it wrong at some step or using some 
> completely different files. Kindly let me know if u can point out any 
> difference from how u do in the above mentioned steps.
>
> Thanks,
> Ryagz
>
>
>
>
> On 06/28/2013 09:18 AM, Harald Pollinger wrote:
>> Hi Ryagz,
>>
>> sadly, my truss doesn't print any version information. Probably I 
>> update mine once from the suacommunity.com site.
>>
>>
>> Do you really run the settings script, or do you source it?
>> It's meant for being sourced, not being run. This is why it doesn't 
>> have the execute flag set after installation.
>>
>> If you run the script, it is executed in a sub shell, it sets the 
>> variables in the sub shell. After the script finished, also the sub 
>> shell finishes and the variables are lost.
>>
>> If you did source it and the variables are not set by the settings 
>> script, it seems your installation wasn't successful. The settings 
>> script should set or modify these variables:
>>
>> SGE_ROOT
>> SGE_CELL
>> SGE_CLUSTER_NAME
>> SGE_QMASTER_PORT
>> SGE_EXECD_PORT
>> MANPATH
>> PATH
>> LD_LIBRARY_PATH
>>
>> If this is not the case, I'd repeat the Qmaster installation.
>>
>> Does this help you?
>>
>> Cheers,
>> Harald
>>
>>
>> On 06/28/2013 07:45 PM, Raghavendra wrote:
>>> Hi Harald,
>>>
>>> Thanks for trying it out for me. Now i am confident that we could make
>>> it work some how.
>>>
>>> Yea i have followed the installation guide and have disabled data
>>> execution prevention.
>>>
>>> When i run the following in cmd prompt, i get:
>>>
>>> C:\Users\vizexperts>wmic OS Get DataExecutionPrevention_SupportPolicy
>>> DataExecutionPrevention_SupportPolicy
>>> 0
>>>
>>> from an online resource, the result 0 means DEP set to always off.
>>>
>>> So DEP should not be the issue. One more observation i have is truss is
>>> giving the same error for wat ever command i execute. I Simply tried
>>>
>>> #truss ls
>>> tracing pid 1051
>>> getdata() getdata returned 0
>>> process killed by signal 9
>>>
>>> I am not sure if truss is not setup properly or if i am having some
>>> issues at SUA level only.
>>>
>>> I have installed SUA as decribed in the document and installed various
>>> packages using debian-interix link http://debian-interix.net/ provided
>>> by bill.
>>>
>>> Another issue - when i just run default/common/settings.sh and then run
>>> qconf -sconf, the first error i get is missing environment variable for
>>> SGE_QMASTER_PORT and then for SGE_EXECD_PORT after i set those
>>> environment variables in the shell i then get this libssl error. So, 
>>> are
>>> these env variable missing errors expected? because i didnt face 
>>> them in
>>> linux.
>>>
>>>
>>> Thanks,
>>> Ryagz
>>>
>>>
>>>
>>> On 06/28/2013 07:25 AM, Harald Pollinger wrote:
>>>> Hi Ryagz,
>>>>
>>>> I just tested the win32-x86 binaries from the UGE 8.1.4 demo package.
>>>>
>>>> I ran
>>>> # truss bin/win32-x86/qconf -sconf
>>>> and got about 50 lines of output and an "exit(1)" at the end.
>>>>
>>>> Then I installed Qmaster on a Linux host and the Execd on the Windows
>>>> host, unset LD_LIBRARY_PATH and ran "qconf -sconf" again in the truss
>>>> and got a lot of output including:
>>>>
>>>> open("/work/clusters/V814/lib/win32-x86/libssl.so", 0x1) open 
>>>> returned 3
>>>> read(3, 0x19FDBB0, 4096) read returned 4096 0x1000
>>>> close(3) close returned 0
>>>>
>>>> I.e. the libssl.so was successfully opened, i.e. in principle it 
>>>> works.
>>>>
>>>>
>>>>
>>>> You got "process killed by signal 9" in your output - did you disable
>>>> the Data Execution Prevention?
>>>>
>>>> See
>>>> http://social.technet.microsoft.com/Forums/windows/en-US/0f4f1fa2-bef2-483e-a00f-9c5224e1f403/turn-off-dep 
>>>>
>>>>
>>>>
>>>> Like Bill wrote, there are a lot of things to take care of, following
>>>> the installation guide is really necessary here...
>>>>
>>>> Cheers,
>>>> Harald
>>>>
>>>>
>>>> On 06/28/2013 01:10 PM, Raghavendra wrote:
>>>>> Hi Harald,
>>>>>
>>>>> Thanks for the reply.
>>>>>
>>>>> if its looking for libraries in ../../lib/win32-x86 then they are
>>>>> readily available there. Using truss command i am getting following
>>>>> error:
>>>>>
>>>>> -sh-3.1$ truss qconf -sconf
>>>>> tracing pid 1051
>>>>> getdata() getdata returned 0
>>>>> process killed by signal 9
>>>>>
>>>>> Any idea?
>>>>>
>>>>> Ryagz
>>>>>
>>>>> On 06/27/2013 03:22 PM, Harald Pollinger wrote:
>>>>>> The SFU/SUA binaries are built with the -rpath option of the linker.
>>>>>> The -rpath option tells the loader to search for the libraries 
>>>>>> not in
>>>>>> the LD_LIBRARY_PATH, but in the specified path, which is
>>>>>> "../../lib/win32-x86" from the location of the binary you want to 
>>>>>> run.
>>>>>>
>>>>>> So if the qconf binary is in the "$SGE_ROOT/bin/win32-x86" 
>>>>>> directory,
>>>>>> it should find the ssl library in the "../../lib/win32-x86" path.
>>>>>>
>>>>>> If you have installed truss, then run
>>>>>> # truss qconf -sconf
>>>>>> to see which ssl lib the loader really wants to open.
>>>>>>
>>>>>> Cheers,
>>>>>> Harald
>>>>>>
>>>>>> On 06/28/2013 01:41 AM, Raghavendra wrote:
>>>>>>> Hi Bill,
>>>>>>>
>>>>>>> Thanks for your effort in searching for me. Debian for interix is
>>>>>>> working, i was able to install openssh for sua. But one problem 
>>>>>>> still
>>>>>>> exists.
>>>>>>>
>>>>>>> When i run qconf -sh in the sua command prompt after loading
>>>>>>> settings.sh
>>>>>>> file. I am getting the following error:
>>>>>>>
>>>>>>> error: commlib error: can't open ssl library (Unable to open the
>>>>>>> OpenSSL
>>>>>>> library
>>>>>>> .  Please make sure libssl is accessible from your shared library
>>>>>>> path.)
>>>>>>> ERROR: unable to send message to qmaster using port 6444 on host
>>>>>>> "raghavendra":
>>>>>>> framework is not initialized
>>>>>>>
>>>>>>>  From the error i understood its trying to load library libssl. I
>>>>>>> searched and found the libraries libssl.so and libssl.so.1.0.0 are
>>>>>>> available in $SGE_ROOT\lib\win32-x86 directory. I even tried adding
>>>>>>> this
>>>>>>> directory to $LD_LIBRARY_PATH but didnt work.
>>>>>>> On digging on the internet i found cl_com_ssl_crypto_handle =
>>>>>>> sge_dlopen
>>>>>>> ("libssl", LIBSSL_VER); dlopen is failing.
>>>>>>>
>>>>>>> What would be the possible reasons? Incompatible libssl libraries?
>>>>>>> These libraries have been taken from univa win32 tar file.
>>>>>>>
>>>>>>> Kindly help me to resolve the issues.
>>>>>>>
>>>>>>> Thanks,
>>>>>>> Ryagz
>>>>>>> On 06/26/2013 08:33 AM, Bill Bryce wrote:
>>>>>>>> Another possible useful link Ryagaz....Debian has an Interix 
>>>>>>>> port for
>>>>>>>> tools that include OpenSSH.  You can find it here:
>>>>>>>> http://debian-interix.net/
>>>>>>>>
>>>>>>>> Regards,
>>>>>>>>
>>>>>>>> Bill.
>>>>>>>>
>>>>>>>> On 2013-06-26, at 11:16 AM, Raghavendra wrote:
>>>>>>>>
>>>>>>>>> Hi,
>>>>>>>>>
>>>>>>>>> I found that scalable logic guys have implemented support of 
>>>>>>>>> cygwin
>>>>>>>>> from the link bellow:
>>>>>>>>>
>>>>>>>>> http://blogs.scalablelogic.com/2012/06/grid-engine-cygwin-port.html 
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> But i am not able to find binaries for the above anywhere over 
>>>>>>>>> the
>>>>>>>>> net. I tried contacting scalable logic guys about this, but there
>>>>>>>>> hasn't been any reply from them. Is any one aware of how to 
>>>>>>>>> use the
>>>>>>>>> above mentioned in cygwin?
>>>>>>>>>
>>>>>>>>> And Is scalable logic  still an active group? If so can any one
>>>>>>>>> provide us their contact information other than
>>>>>>>>> info at scalablelogic.com <mailto:info at scalablelogic.com>. 
>>>>>>>>> Because we
>>>>>>>>> are interested in opting for their commercial support program if
>>>>>>>>> cygwin implementation works for windows 7 64 bit version.
>>>>>>>>>
>>>>>>>>> Regards,
>>>>>>>>> Ryagz
>>>>>>>>>
>>>>>>>>> _______________________________________________
>>>>>>>>> users mailing list
>>>>>>>>> users at gridengine.org <mailto:users at gridengine.org>
>>>>>>>>> https://gridengine.org/mailman/listinfo/users
>>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> users mailing list
>>>>>>> users at gridengine.org
>>>>>>> https://gridengine.org/mailman/listinfo/users
>>>>>>>
>>>>>> _______________________________________________
>>>>>> users mailing list
>>>>>> users at gridengine.org
>>>>>> https://gridengine.org/mailman/listinfo/users
>>>>>>
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> users mailing list
>>>>> users at gridengine.org
>>>>> https://gridengine.org/mailman/listinfo/users
>>>> _______________________________________________
>>>> users mailing list
>>>> users at gridengine.org
>>>> https://gridengine.org/mailman/listinfo/users
>>>>
>>>>
>>>
>>> _______________________________________________
>>> users mailing list
>>> users at gridengine.org
>>> https://gridengine.org/mailman/listinfo/users
>> _______________________________________________
>> users mailing list
>> users at gridengine.org
>> https://gridengine.org/mailman/listinfo/users
>>
>>
>
>
>
> _______________________________________________
> users mailing list
> users at gridengine.org
> https://gridengine.org/mailman/listinfo/users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://gridengine.org/pipermail/users/attachments/20130701/6bae1416/attachment-0001.html>


More information about the users mailing list