[Sidefx-houdini-list] HQueue 'paster' process at 100%

Nick Nimble nick at nicknimble.com
Tue Jan 29 11:48:33 EST 2013


For the mailing list, I had a discussion with Rob through the bug-db and 
the solution that worked for me, though be it a bit rough, is to wipe 
the hq database. I lost all the running jobs, but I rather did that than 
not rendering.

Commands to do this on linux:

  - Shut down the server, clients and any web browser windows connected 
to HQueue.
- On the server machine, temporarily rename the database file in 
/opt/hqueue/db/hqserver.db
- Then recreate the database with these 2 commands:
      cd /opt/hqueue
      LD_LIBRARY_PATH=/opt/hqueue/lib:${LD_LIBRARY_PATH} ./bin/paster 
setup-app hqserver.ini
     - Start the HQueue server.
     - Restart the HQueue clients or re-add them through the HQueue web 
interface.

Cheers,
Nick, be Nimble
+31 622 4243 94 | www.nicknimble.com | www.facebook.com/nicknimble

On 1/29/13 5:23 PM, Robert Vinluan wrote:
> Hi Nick,
>
> The 'paster' process *is* the HQueue server so if you kill paster, you 
> shutdown the server.
>
> You have a couple of options:
> 1) Recover the system.
> - To do that, you need to first shutdown the server, shutdown all the 
> clients and close all web browser windows that are viewing the HQueue 
> web interface.
> - Then restart only the server.  The paster process will initially 
> peak at 100% as it tries to recover and load all the active jobs from 
> the database.  Wait until the process has settled down and its 
> percentage has dropped.
> - Open a web browser window and navigate to HQueue.
> - Check the box beside the 6000 frame job and click Cancel.  Once the 
> job has been moved to the bottom table, then check it again and click 
> on Delete.
> - Now that the large job has been cleared, restart the HQueue clients.
> 2) Start with a clean system and new database.
> - Shut down the server, clients and any web browser windows connected 
> to HQueue.
> - On the server machine, temporarily rename the database file in 
> /opt/hqueue/db/hqserver.db
> - Then recreate the database with these 2 commands:
>     cd /opt/hqueue
>     LD_LIBRARY_PATH=/opt/hqueue/lib:${LD_LIBRARY_PATH} ./bin/paster 
> setup-app hqserver.ini
> - Start the HQueue server.
> - Restart the HQueue clients or re-add them through the HQueue web 
> interface.
>
> Once your system is up and running again, I suggest breaking up your 
> 6000 frame job into batches of say 300, or 500 frames. HQueue will be 
> able to better handle the job when it is broken up in pieces.
>
> Anyway, I hope this helps.
>
> Cheers,
> Rob
>
> On 13-01-29 10:53 AM, Nick Nimble wrote:
>> Hi,
>>
>> I hope there is someone who can help me out. I'm in the middle of a 
>> crunch and HQueue stopped working.
>>
>> If I login on te server that runs HQ then in 'top' it show a process 
>> 'paster' at 100%. I've tried to kill the process and do a reboot, but 
>> as soon as the machine comes back online the paster process is there. 
>> The paster process seems to be directly related to the hqserver 
>> because when I stop the server the process is gone, when I start 
>> hqserver its back.
>>
>> Do you have any suggestions how I can resolve this? It is ok to lose 
>> the current render job. I just want the server to respond and except 
>> new jobs. I run Ubuntu 12.04lts on all machines.
>>
>> Can I delete any cache files of some sort? Last night I submitted a 
>> job of 6000 frames, 2 output drivers in stereo and due to an mistake 
>> on my end all the frames we're rendering black and within 30 -45sec a 
>> frame. To me it seems likely that something overrun, like the db?
>>
>> I've tried to rename the database file and create a new one with 
>> ./bin/paster setup-app hqserver.ini but it returns: 
>> /opt/hqueue/bin/python2.6-bin: error while loading shared libraries: 
>> libpython2.6.so.1.0: cannot open shared object file: No Such file or 
>> directory
>>
>> If you feel like saving the day, this is your opportunity ;-)
>>
>




More information about the Sidefx-houdini-list mailing list