Thanks! Becky -- Becky Ligon HPC Admin Staff PVFS/OrangeFS Developer Clemson University/Omnibond.com OrangeFS Support 864-650-4065
> Object Storage Device > > http://en.wikipedia.org/wiki/Object_storage_device > http://www.open-osd.org/bin/view > > A standardized object server. > > Scott > > On Jun 2, 2011, at 1:24 PM, Becky Ligon wrote: > >> What is OSD? >> >> Becky >> -- >> Becky Ligon >> HPC Admin Staff >> PVFS/OrangeFS Developer >> Clemson University/Omnibond.com OrangeFS Support >> 864-650-4065 >> >>> Hi Rob, >>> >>> I can answer for Cengiz. Pete Wyckoff and the OSC folk had done some >>> work >>> getting PVFS to work with OSD. That was with PVFS 2.7.1. We are >>> trying >>> to bring it up to date with 2.8.4. Since OSDs do not have precreate >>> support, we need a way to disable that functionality. It looks like we >>> are going to have dig into the state machines and follow Michael's >>> suggestions. >>> >>> John. >>> >>> On Jun 2, 2011, at 10:56 AM, Rob Ross wrote: >>> >>>> Hi, >>>> >>>> In short Cengiz, you're going to have to do a little work if you want >>>> to >>>> turn off the precreate functionality and go back to the old way of >>>> allocating metadata and data objects. That code isn't being tested or >>>> supported any more; the new way of doing things is simply better for >>>> most users. Another option would be to use an older version of PVFS >>>> that >>>> does not have this feature. >>>> >>>> Rob >>>> >>>> On Jun 2, 2011, at 9:51 AM, Michael Moore wrote: >>>> >>>>> I checked a little closer in the create state machine. If I'm reading >>>>> it correctly, the only way to only get local file handles it to have >>>>> the stuffed option on. Otherwise, remote file handles will be >>>>> allocated >>>>> and the only way those currently get populated is from precreate >>>>> pools. >>>>> Of course testing the configuration is the only definitive way to >>>>> know >>>>> where and if it breaks. >>>>> >>>>> To change the reliance on precreate behavior my thought would be to >>>>> change the state machines that rely on >>>>> job_precreate_pool_get_handles() >>>>> to have a different way to get handles. In other words, based on some >>>>> flag have the state machine take one of two paths 1) get handles >>>>> normally with precreate 2) the old route. Now, I am not familiar with >>>>> the 'old route' so I don't know how well it would fit. >>>>> >>>>> Michael >>>>> >>>>> On Thu, Jun 2, 2011 at 8:39 AM, Phil Carns <[email protected]> wrote: >>>>> I share Michael's curiosity as well, but I think it might be possible >>>>> to disable the precreate functionality using config file options if >>>>> you >>>>> really need to (not recommended): >>>>> >>>>> Set the FileStuffing option to "no" to prevent the file system from >>>>> using the file stuffing optimization that relies on precreated >>>>> objects: >>>>> >>>>> blockedhttp://www.pvfs.org/cvs/pvfs-2-8-branch-docs/doc//pvfs-config-options.php#FileStuffing >>>>> >>>>> And then also set the PrecreateLowThreshold to 0 to prevent the >>>>> servers >>>>> from trying to fill up their respective pools of precreated objects >>>>> on >>>>> startup: >>>>> >>>>> blockedhttp://www.pvfs.org/cvs/pvfs-2-8-branch-docs/doc//pvfs-config-options.php#PrecreateLowThreshold >>>>> >>>>> I have never tried that particular configuration so I can't say for >>>>> sure if it will work or not. >>>>> >>>>> -Phil >>>>> >>>>> >>>>> On 06/01/2011 09:52 PM, Michael Moore wrote: >>>>>> >>>>>> Hi Cengiz, >>>>>> >>>>>> I'm curious, what's the use case for needing the precreate >>>>>> functionality disabled? Migration from an older version? >>>>>> >>>>>> Thanks, >>>>>> Michael >>>>>> >>>>>> On Wed, Jun 1, 2011 at 3:01 PM, Karakoyunlu, Cengiz >>>>>> <[email protected]> wrote: >>>>>> Is it possible to disable the precreate functions in orangefs-2.8.4 >>>>>> and use the traditional methods existed in previous distributions to >>>>>> start server, perform i/o operations etc.? >>>>>> >>>>>> Thank you, >>>>>> >>>>>> Cengiz >>>>>> >>>>>> _______________________________________________ >>>>>> Pvfs2-developers mailing list >>>>>> [email protected] >>>>>> blockedhttp://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers >>>>>> >>>>>> >>>>>> >>>>>> _______________________________________________ >>>>>> Pvfs2-developers mailing list >>>>>> [email protected] >>>>>> blockedhttp://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers >>>>> >>>>> >>>>> _______________________________________________ >>>>> Pvfs2-developers mailing list >>>>> [email protected] >>>>> blockedhttp://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers >>>>> >>>>> >>>>> _______________________________________________ >>>>> Pvfs2-developers mailing list >>>>> [email protected] >>>>> blockedhttp://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers >>>> >>>> _______________________________________________ >>>> Pvfs2-developers mailing list >>>> [email protected] >>>> blockedhttp://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers >>> >>> >>> _______________________________________________ >>> Pvfs2-developers mailing list >>> [email protected] >>> blockedhttp://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers >>> >> >> >> _______________________________________________ >> Pvfs2-developers mailing list >> [email protected] >> blockedhttp://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers >> > > _______________________________________________ Pvfs2-developers mailing list [email protected] http://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers
