Re: UUIDs eventually choke as primary key
Well, I don't seem to be able to get around this. I'm going to have to
switch away from using a UUID as a primary key, which fortunately doesn't
seem like too much work.
On Thursday, March 1, 2018 at 8:53:13 AM UTC-8, M Mackey wrote:
>
> I added some debug support, and there are only two classes called UUID
> when this fails:
> uuid UUID:
> psycopg2.extensions UUID: '.../python3.6/site-packages/psycopg2/extensions.py'>
>
> The psycopg extension is a converter, and it seems highly unlikely that it
> was put into place as the value, and I doubt it would report it's __str__
> as the UUID string, so it does not appear to be a case of mistaken
> identity. Which makes the failure of "not instance" to catch the UUID after
> a while all the more weird...
>
> I'm pretty baffled on this one. Am I really the only one running into this?
>
> M
>
> On Friday, February 23, 2018 at 1:17:34 PM UTC-8, M Mackey wrote:
>>
>> Not to belabor the point, but this is the part that I find weird. This is
>> down in the "During handling of the above exception ('UUID' object has no
>> attribute 'replace'), another exception occurred" section.
>>
>> /usr/local/venvs/attrack/lib/python3.6/site-packages/django/db/models/fields/__init__.py
>>
>> in to_python
>> if value is not None and not isinstance(value, uuid.UUID):
>> <<
>> 1) CHECKS FOR NOT UUID
>> try:
>> return uuid.UUID(value) <<- 2) Pretty sure first
>> exception is in here
>> except (AttributeError, ValueError):
>> raise exceptions.ValidationError(
>> self.error_messages['invalid'],
>> code='invalid',
>> params={'value': value}, <<-- Traceback says
>> new exception here
>> )
>> return value
>>
>> Local vars
>> Variable Value
>> self
>> value UUID('338fa43f-3e07-4e83-8525-d195d2fc64d4') << 3)
>> REPORTS AS UUID
>>
>> So, if 3) is right and 'value' is a UUID, 1) should keep us from getting
>> to 2). But we get there.
>>
>> On Friday, February 23, 2018 at 12:59:28 PM UTC-8, M Mackey wrote:
>>>
>>> Adding details I seem to have left out. The id is defined like this:
>>> id = models.UUIDField(primary_key=True, default=uuid.uuid1,
>>> editable=False)
>>> and I'm running on PostgreSQL.
>>> I'm pretty sure I never create a row without already having the key.
>>>
>>>
--
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+unsubscr...@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/038daba7-fa72-4c17-8944-3c15ffb6a8f6%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Re: UUIDs eventually choke as primary key
I added some debug support, and there are only two classes called UUID when
this fails:
uuid UUID:
psycopg2.extensions UUID:
The psycopg extension is a converter, and it seems highly unlikely that it
was put into place as the value, and I doubt it would report it's __str__
as the UUID string, so it does not appear to be a case of mistaken
identity. Which makes the failure of "not instance" to catch the UUID after
a while all the more weird...
I'm pretty baffled on this one. Am I really the only one running into this?
M
On Friday, February 23, 2018 at 1:17:34 PM UTC-8, M Mackey wrote:
>
> Not to belabor the point, but this is the part that I find weird. This is
> down in the "During handling of the above exception ('UUID' object has no
> attribute 'replace'), another exception occurred" section.
>
> /usr/local/venvs/attrack/lib/python3.6/site-packages/django/db/models/fields/__init__.py
>
> in to_python
> if value is not None and not isinstance(value, uuid.UUID): <<
> 1) CHECKS FOR NOT UUID
> try:
> return uuid.UUID(value) <<- 2) Pretty sure first
> exception is in here
> except (AttributeError, ValueError):
> raise exceptions.ValidationError(
> self.error_messages['invalid'],
> code='invalid',
> params={'value': value}, <<-- Traceback says new
> exception here
> )
> return value
>
> Local vars
> Variable Value
> self
> value UUID('338fa43f-3e07-4e83-8525-d195d2fc64d4') << 3) REPORTS
> AS UUID
>
> So, if 3) is right and 'value' is a UUID, 1) should keep us from getting
> to 2). But we get there.
>
> On Friday, February 23, 2018 at 12:59:28 PM UTC-8, M Mackey wrote:
>>
>> Adding details I seem to have left out. The id is defined like this:
>> id = models.UUIDField(primary_key=True, default=uuid.uuid1,
>> editable=False)
>> and I'm running on PostgreSQL.
>> I'm pretty sure I never create a row without already having the key.
>>
>>
--
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+unsubscr...@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/fd5fb988-78ae-40b7-ab5f-ca2d361b7d56%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Re: UUIDs eventually choke as primary key
Not to belabor the point, but this is the part that I find weird. This is
down in the "During handling of the above exception ('UUID' object has no
attribute 'replace'), another exception occurred" section.
/usr/local/venvs/attrack/lib/python3.6/site-packages/django/db/models/fields/__init__.py
in to_python
if value is not None and not isinstance(value, uuid.UUID): <<
1) CHECKS FOR NOT UUID
try:
return uuid.UUID(value) <<- 2) Pretty sure first
exception is in here
except (AttributeError, ValueError):
raise exceptions.ValidationError(
self.error_messages['invalid'],
code='invalid',
params={'value': value}, <<-- Traceback says new
exception here
)
return value
Local vars
Variable Value
self
value UUID('338fa43f-3e07-4e83-8525-d195d2fc64d4') << 3) REPORTS
AS UUID
So, if 3) is right and 'value' is a UUID, 1) should keep us from getting to
2). But we get there.
On Friday, February 23, 2018 at 12:59:28 PM UTC-8, M Mackey wrote:
>
> Adding details I seem to have left out. The id is defined like this:
> id = models.UUIDField(primary_key=True, default=uuid.uuid1, editable=False)
> and I'm running on PostgreSQL.
> I'm pretty sure I never create a row without already having the key.
>
>
--
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+unsubscr...@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/ea0820eb-8cf4-478e-ba4a-8535c420fce0%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Re: UUIDs eventually choke as primary key
Adding details I seem to have left out. The id is defined like this: id = models.UUIDField(primary_key=True, default=uuid.uuid1, editable=False) and I'm running on PostgreSQL. I'm pretty sure I never create a row without already having the key. On Monday, February 12, 2018 at 1:00:53 PM UTC-8, M Mackey wrote: > > I've run into a strange issue with using a UUID as primary key, and I'm > hoping we can either verify this is a bug, or figure out what I've done > wrong. > > I've got a core model object with a UUID for it's primary key. (Generated > external to this system, thus using that for when additional information > comes in.) And there are other model objects referencing it. Normally > everything runs fine. But after a while, web pages that have been running > fine start telling me that the UUID isn't valid (even though it is). (Sorry > I don't have a copy of the error at this moment, I'll post here when it > happens again.) I know it's not a simple coding error, because I can simply > restart Apache and everything is fine again. And these aren't POST > responses or anything complex like that. It occurs on a simple listing page. > > When I dig into the exceptions, I see something that seems contradictory. > It looks like the UUID constructor has been called with an instance of a > UUID (it says type UUID doesn't have method 'replace'). But it doesn't seem > like that should be possible because the code that calls that constructor > first checks the type. (See django/db/models/fields/__init__.py - > UUIDField.to_python) > So it's a double head scratcher. Once this problem crops up, I have to > restart the web server (but then it's okay for a while). And the errors it > gives me don't make much sense. > I don't know what kind of caching problem it could be - I haven't enabled > any of the available caching systems. Also doesn't seem to crop up (at > least not as often) in the built in dev test server. > > Anyone seen this, or have any ideas? > > Thanks, > M > -- 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+unsubscr...@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/dac6114e-b1cb-4b00-b313-8357eaf59ae7%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: UUIDs eventually choke as primary key
My notes don't say I did that, and I tried a new env and sys.path still shows /usr/local/lib/python3.6. Maybe that's just a red herring. But I don't have any other good explanation for the UUID getting passed to the UUID __init__. On Friday, February 23, 2018 at 9:12:33 AM UTC-8, Tom Evans wrote: > > On Fri, Feb 23, 2018 at 4:20 PM, M Mackey <mm...@mac.com > > wrote: > > I have noticed in the python > > path that there are two paths to .../lib/python3.6. One from my > virtualenv, > > and one at /usr/local/. Not sure where to clear that up, since I don't > > believe I've got my apache env set up to pull from both places. > > That would happen if you created your virtualenv with > --use-site-packages, it, er, puts the path to the site packages in the > pythonpath. > > Cheers > > Tom > -- 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+unsubscr...@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/0f153caa-93f0-4851-ade0-c1e1aee0af81%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: UUIDs eventually choke as primary key
The thing that gets me is this part:
.../django/db/models/fields/__init__.py in to_python
> return uuid.UUID(value) ...
> /usr/local/lib/python3.6/uuid.py in __init__
> hex = hex.replace('urn:', '').replace('uuid:', '') ...
>
> During handling of the above exception ('UUID' object has no attribute
> 'replace'), another exception occurred:
>
If you dig into the code, it looks like (based on the exception) a UUID is
being passed into the UUID constructor. But the lines before that
constructor call happens check to see if it's a UUID. I'm wondering if I'm
somehow getting two different UUID classes. I have noticed in the python
path that there are two paths to .../lib/python3.6. One from my virtualenv,
and one at /usr/local/. Not sure where to clear that up, since I don't
believe I've got my apache env set up to pull from both places.
--
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+unsubscr...@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/d34e8538-dede-4540-843f-6c30255be951%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Re: UUIDs eventually choke as primary key
Got it to happen inside the admin interface. It looks like, at least in
this case, it's some kind of cache. I don't see the cache stuff on the
user's listing page. I'll see if I can put a condensed version of the error
here:
ValidationError at /admin//log/1418765/change/
["'f45fecf0-2cad-4c66-a67f-41665732086c' is not a valid UUID."]
Error during template rendering
In
template.../python3.6/site-packages/django/contrib/admin/templates/admin/base.html,
error at line 0
'%(value)s' is not a valid UUID.
1 {% load i18n static %}
2 {% get_current_language as LANGUAGE_CODE %}{% get_current_language_bidi
as LANGUAGE_BIDI %}
3
4
5 {% block title %}{% endblock %}
6
7 {% block extrastyle %}{% endblock %}
8 {% if LANGUAGE_BIDI %}{%
endif %}
9 {% block extrahead %}{% endblock %}
10 {% block responsive %}
Traceback
.../django/db/models/fields/related_descriptors.py in __get__
rel_obj = self.field.get_cached_value(instance) ...
.../django/db/models/fields/mixins.py in get_cached_value
return instance._state.fields_cache[cache_name] ...
During handling of the above exception ('package'), another exception
occurred:
.../django/db/models/fields/__init__.py in to_python
return uuid.UUID(value) ...
/usr/local/lib/python3.6/uuid.py in __init__
hex = hex.replace('urn:', '').replace('uuid:', '') ...
During handling of the above exception ('UUID' object has no attribute
'replace'), another exception occurred:
.../django/core/handlers/exception.py in inner
On Monday, February 12, 2018 at 1:00:53 PM UTC-8, M Mackey wrote:
>
> I've run into a strange issue with using a UUID as primary key, and I'm
> hoping we can either verify this is a bug, or figure out what I've done
> wrong.
>
> I've got a core model object with a UUID for it's primary key. (Generated
> external to this system, thus using that for when additional information
> comes in.) And there are other model objects referencing it. Normally
> everything runs fine. But after a while, web pages that have been running
> fine start telling me that the UUID isn't valid (even though it is). (Sorry
> I don't have a copy of the error at this moment, I'll post here when it
> happens again.) I know it's not a simple coding error, because I can simply
> restart Apache and everything is fine again. And these aren't POST
> responses or anything complex like that. It occurs on a simple listing page.
>
> When I dig into the exceptions, I see something that seems contradictory.
> It looks like the UUID constructor has been called with an instance of a
> UUID (it says type UUID doesn't have method 'replace'). But it doesn't seem
> like that should be possible because the code that calls that constructor
> first checks the type. (See django/db/models/fields/__init__.py -
> UUIDField.to_python)
> So it's a double head scratcher. Once this problem crops up, I have to
> restart the web server (but then it's okay for a while). And the errors it
> gives me don't make much sense.
> I don't know what kind of caching problem it could be - I haven't enabled
> any of the available caching systems. Also doesn't seem to crop up (at
> least not as often) in the built in dev test server.
>
> Anyone seen this, or have any ideas?
>
> Thanks,
> M
>
--
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+unsubscr...@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/a3cd1433-f44c-4e10-b005-9b177244833a%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
UUIDs eventually choke as primary key
I've run into a strange issue with using a UUID as primary key, and I'm hoping we can either verify this is a bug, or figure out what I've done wrong. I've got a core model object with a UUID for it's primary key. (Generated external to this system, thus using that for when additional information comes in.) And there are other model objects referencing it. Normally everything runs fine. But after a while, web pages that have been running fine start telling me that the UUID isn't valid (even though it is). (Sorry I don't have a copy of the error at this moment, I'll post here when it happens again.) I know it's not a simple coding error, because I can simply restart Apache and everything is fine again. And these aren't POST responses or anything complex like that. It occurs on a simple listing page. When I dig into the exceptions, I see something that seems contradictory. It looks like the UUID constructor has been called with an instance of a UUID (it says type UUID doesn't have method 'replace'). But it doesn't seem like that should be possible because the code that calls that constructor first checks the type. (See django/db/models/fields/__init__.py - UUIDField.to_python) So it's a double head scratcher. Once this problem crops up, I have to restart the web server (but then it's okay for a while). And the errors it gives me don't make much sense. I don't know what kind of caching problem it could be - I haven't enabled any of the available caching systems. Also doesn't seem to crop up (at least not as often) in the built in dev test server. Anyone seen this, or have any ideas? Thanks, M -- 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+unsubscr...@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/d283cde8-76ff-4218-a504-4a92508ae2a2%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
