Fred
Spot on!Saving to the production database didn't work. It couldn't get
past the atomicity protection in Postgres. I gave up and did it manually
for the fellow who had got himself into that pickle. He loves me now.
It's embarrassing!
We also adjusted the user docs to say the staging server gets blown away
frequently - use it at your peril!
I do like your DRAFT notion.
Thanks
Mike
On 3/01/2017 11:52 AM, Fred Stluka wrote:
> Mike,
>
> On 11/11/16 8:00 AM, Mike Dewhirst wrote:
> > ... My solution is to tell the users to avoid doing real work on the
> > staging server. It exists only to check out new features. ...
>
> I have a similar issue with users of my TEST vs PROD web sites. They
> develop experimental data in TEST and then want me to move it to
> PROD once they have it ready.
>
> So far, I've been able to accommodate them by dumping part or all
> of the TEST DB and importing it into the PROD DB using native tools
> that come with the DBMS (MySQL in this case).
>
> But, the best solution, I think, is to add a "Draft" field to the models,
> so that they can perform their experimental data changes in PROD
> flagging them as draft, and then clear the draft flag when they are
> ready for other users to see the new data.
>
> This may require an access control system that allows privileged
> users to mark data as draft, and to specify which other users can
> see/edit the draft data, etc. May also need multiple different drafts,
> so you get into version tracking, etc. Not simple, but may be the
> best answer to your problem.
>
> --Fred
> ------------------------------------------------------------------------
> Fred Stluka -- mailto:fred@bristle.com -- http://bristle.com/~fred/
> Bristle Software, Inc -- http://bristle.com -- Glad to be of service!
> Open Source: Without walls and fences, we need no Windows or Gates.
> ------------------------------------------------------------------------
>
> On 11/11/16 8:00 AM, Mike Dewhirst wrote:
>> To finish this thread, I have decided to abandon the attempt. My
>> solution is to tell the users to avoid doing real work on the staging
>> server. It exists only to check out new features.
>>
>> I feel better now.
>>
>> Thanks Melvyn and Tim
>>
>> Mike
>>
>> On 10/11/2016 11:08 AM, Mike Dewhirst wrote:
>>> On 10/11/2016 12:06 AM, Tim Graham wrote:
>>>> I see. Although I would be interested to see what you come up with,
>>> @Tim
>>> It is getting there so if you are really interested I'll post it
>>> somewhere when it is working. I'm getting router errors at the
>>> moment which I might be able to workaround by permitting null in a
>>> FK fields during obj.save(using=db). Might ask some questions on
>>> this later. I don't really want to permit null in this particular case.
>>>
>>>> my initial reaction is that we wouldn't accept a patch for Django
>>>> to do this. It seems too complex and brittle.
>>>
>>> Having done some work on this I agree with you but disagree with
>>> Melvyn.
>>>
>>> The task is complex and would be almost impossible to generalise.
>>> For example, from a number of substances the user may have done
>>> serious work on only a few and won't want the others transferred.
>>>
>>> So I'm building a facility to transfer one substance at a time. If
>>> there is a future requirement for bulk transfer I can iterate.
>>>
>>> The transferred substance might be a mixture in which case I have to
>>> collect the ingredient substances plus the m2m through records which
>>> are effectively the recipe. Some of the ingredients might themselves
>>> be mixtures. Not worrying about that at the moment.
>>>
>>> For each substance there are 6 OneToMany, up to 15 OneToOne and two
>>> m2m relationships plus a OneToOne where the sibling has lots of its
>>> own complex relationships including a recursive OneToOne to form a
>>> daisy-chain of history. I'm skipping that one for the moment.
>>>
>>>> I agree with Melvyn
>>>
>>> See below
>>>
>>>> that it seems unusual to use Django's multi-database facilities to
>>>> transfer data between staging and production.
>>>>
>>>> On Wednesday, November 9, 2016 at 4:51:39 AM UTC-5, Melvyn Sopacua
>>>> wrote:
>>>>
>>>> On Wednesday 09 November 2016 13:59:35 Mike Dewhirst wrote:
>>>> > On 9/11/2016 9:13 AM, Tim Graham wrote:
>>>> > > Interesting idea. I'd be interested to here more about the use
>>>> case.
>>>> >
>>>> > Simple. I permitted my users to "play" on the staging site and
>>>> now one
>>>> > of them wants to transfer their work to the production site.
>>>>
>>>> Your approach is wrong. You think of this as one django project
>>>> and
>>>> trying to cheat under the hood, while in fact they are two
>>>> different
>>>> projects. And by design, staging is not always structually
>>>> identical to
>>>> production.
>>>>
>>>
>>> @Melvyn
>>>
>>> I was wrong to think saveasnew might do the trick.
>>>
>>> However, migration is fully implemented so I can easily abort any
>>> transfer if the structures are not identical. As it happens, there
>>> are only two databases at the moment. Staging and production and
>>> while they are not always structurally identical they are currently
>>> the same during testing.
>>>
>>>>
>>>> Your best approach is to use serialization with natural keys
>>>
>>> Maybe but the complexity and a couple of business rules persuade me
>>> to write code to do it "manually" so I can unit test it. Also, if I
>>> get it going, the user can select a substance to transfer and
>>> doesn't have to bother me.
>>>
>>> I also realise it is very brittle so it needs to be revised if the
>>> structure changes.
>>>
>>> Thanks for your advice. I appreciate it even if I think I need to
>>> tackle the problem my way.
>>>
>>> Cheers
>>>
>>> Mike
>>>
>>>> and a job
>>>> scheduler. Cron will do, but there's plenty other schemes allowing
>>>> you
>>>> to provide feedback about the job queue and expected time of
>>>> completion.
>>>> -- Melvyn Sopacua
>>>>
>>>> --
>>>> You received this message because you are subscribed to the Google
>>>> Groups "Django users" group.
>>>> To unsubscribe from this group and stop receiving emails from it,
>>>> send an email to django-users+unsubscribe@googlegroups.com
>>>> <mailto:django-users+unsubscribe@googlegroups.com>.
>>>> To post to this group, send email to django-users@googlegroups.com
>>>> <mailto:django-users@googlegroups.com>.
>>>> Visit this group at https://groups.google.com/group/django-users.
>>>> To view this discussion on the web visit
>>>> https://groups.google.com/d/msgid/django-users/ccfa9aef-d1c3-45e0-a894-c07a2701c1aa%40googlegroups.com
>>>> <https://groups.google.com/d/msgid/django-users/ccfa9aef-d1c3-45e0-a894-c07a2701c1aa%40googlegroups.com?utm_medium=email&utm_source=footer>.
>>>>
>>>> For more options, visit https://groups.google.com/d/optout.
>>>
>>
>
> --
> You received this message because you are subscribed to the Google
> Groups "Django users" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to django-users+unsubscribe@googlegroups.com
> <mailto:django-users+unsubscribe@googlegroups.com>.
> To post to this group, send email to django-users@googlegroups.com
> <mailto:django-users@googlegroups.com>.
> Visit this group at https://groups.google.com/group/django-users.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/django-users/f54523dc-9022-77ac-d76a-57588ddea3bc%40bristle.com
> <https://groups.google.com/d/msgid/django-users/f54523dc-9022-77ac-d76a-57588ddea3bc%40bristle.com?utm_medium=email&utm_source=footer>.
> For more options, visit https://groups.google.com/d/optout.
--
You received this message because you are subscribed to the Google Groups "Django users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to django-users+unsubscribe@googlegroups.com.
To post to this group, send email to django-users@googlegroups.com.
Visit this group at https://groups.google.com/group/django-users.
To view this discussion on the web visit https://groups.google.com/d/msgid/django-users/fcb5fcb8-c6c6-4dd6-94eb-65cf7144f7d5%40dewhirst.com.au.
For more options, visit https://groups.google.com/d/optout.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment