[web2py] Re: GAE Hosting Issue

2013-03-06 Thread Christian Foster Howes

Philip,

"I've discovered that if I create an empty database via a direct 
connection to CloudSQL and point the DAL at that"  that statement makes 
me think


web2py *never* sends a "create database" statement to the database (no 
matter what DB you are using).  it expects that the DB server is up and 
running at the configured address, and that the empty database/schema 
(depending on the DB engine the terms change) is already there with 
proper permissions for the user that web2py will try to login with. 
then web2py will issue "create table" statements as needed.  if the 
database/schema does not exist, there will be errors.


cfh

On 3/6/13 0:07 , Philip Kilner wrote:

Hi Christian,

On 06/03/13 07:42, Christian Foster Howes wrote:

the ticket you linked to looks like it is not failing on the connection,
but rather failing on a query.  these 2 line are of interest to me:

   File
"/base/data/home/apps/s~gk-hercules/1.365624770532609305/gluon/dal.py", line
4115, in execute
 return self.log_execute(command.decode('utf8'), *a, **b)
   File
"/base/data/home/apps/s~gk-hercules/1.365624770532609305/gluon/dal.py", line
1703, in log_execute
 ret = self.cursor.execute(*a, **b)



Well spotted - it's spitting out a very wide variety of errors, but I'm
increasingly leaning to thinking it relates to the failed migration.


any chance you are able to open dal.py and put in some debug logging
at those lines an run it again? I'm curious what query/command web2py
is trying to run on the database.  that might open the door to the
missing answer!



I'll do that - I've discovered that if I create an empty database via a
direct connection to CloudSQL and point the DAL at that, the application
is accessible.

Given GAE's recent issues around resources/latency, I'm beginning to
think that this is the real issue, and the other factors are simply
generating spurious errors before it can show me the "real" error.

I'm really anxious to get the migration pickles out of the db that they
are attempting to manage and into GAE's datastore.

I've had to move the app to conventional hosting, as this has been a
huge time-sink, so it will be a day or two before I get back to it.





--

--- 
You received this message because you are subscribed to the Google Groups "web2py-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email 
to web2py+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.




[web2py] Re: GAE Hosting Issue

2013-03-06 Thread Philip Kilner

Hi Christian,

On 06/03/13 07:42, Christian Foster Howes wrote:

the ticket you linked to looks like it is not failing on the connection,
but rather failing on a query.  these 2 line are of interest to me:

   File "/base/data/home/apps/s~gk-hercules/1.365624770532609305/gluon/dal.py", 
line 4115, in execute
 return self.log_execute(command.decode('utf8'), *a, **b)
   File "/base/data/home/apps/s~gk-hercules/1.365624770532609305/gluon/dal.py", 
line 1703, in log_execute
 ret = self.cursor.execute(*a, **b)



Well spotted - it's spitting out a very wide variety of errors, but I'm 
increasingly leaning to thinking it relates to the failed migration.



any chance you are able to open dal.py and put in some debug logging at those 
lines an run it again? I'm curious what query/command web2py is trying to run 
on the database.  that might open the door to the missing answer!



I'll do that - I've discovered that if I create an empty database via a 
direct connection to CloudSQL and point the DAL at that, the application 
is accessible.


Given GAE's recent issues around resources/latency, I'm beginning to 
think that this is the real issue, and the other factors are simply 
generating spurious errors before it can show me the "real" error.


I'm really anxious to get the migration pickles out of the db that they 
are attempting to manage and into GAE's datastore.


I've had to move the app to conventional hosting, as this has been a 
huge time-sink, so it will be a day or two before I get back to it.




--

Regards,

PhilK


e: p...@xfr.co.uk - m: 07775 796 747

'work as if you lived in the early days of a better nation'
- alasdair gray

--

--- 
You received this message because you are subscribed to the Google Groups "web2py-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email 
to web2py+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.




[web2py] Re: GAE Hosting Issue

2013-03-05 Thread Christian Foster Howes
the ticket you linked to looks like it is not failing on the connection, 
but rather failing on a query.  these 2 line are of interest to me:

  File "/base/data/home/apps/s~gk-hercules/1.365624770532609305/gluon/dal.py", 
line 4115, in execute
return self.log_execute(command.decode('utf8'), *a, **b)
  File "/base/data/home/apps/s~gk-hercules/1.365624770532609305/gluon/dal.py", 
line 1703, in log_execute
ret = self.cursor.execute(*a, **b)

any chance you are able to open dal.py and put in some debug logging at those 
lines an run it again? I'm curious what query/command web2py is trying to run 
on the database.  that might open the door to the missing answer!

cfh



On Monday, March 4, 2013 11:07:13 PM UTC-8, Philip Kilner wrote:
>
> Hi Christian, 
>
> On 05/03/13 06:12, Christian Foster Howes wrote: 
> > hrmmaybe it's time for GAE's next code update.  i frequently notice 
> > a degradation in performance just before they announce a new SDK 
> version. 
> > 
>
> Interesting observation - will look out for that. 
>
> > i'm not seeing anything out of the ordinary on my GAE apps right now, 
> > but i do agree that the GAE team is not as transparent as i would like. 
> > 
> > i would suggest that you read the logs very carefully - there are lots 
> > of different deadlineexceeded exceptions (literally the same class names 
> > with different paths) and try adding extra logging to see where things 
> > are getting stuck. 
> > 
>
> Actually, I'm not getting past a db connection so there is nothing in my 
> application code to log, and the logged errors accessible in the console 
> are very, very varied [1]. 
>
> > i don't know if web2py does automatic migrations on cloudsql, and in 
> > production i always do my SQL by handbut i'm old school and 
> pessimistic. 
> > 
>
> It does - I did have a failed migration (how I wish their CloudSQL 
> offering was based on Postgres!), but the app was buzzing away happily 
> with the help of "fake_migrate_all" for a good while after that. 
>
> The migration issues is messier than I'd like it to be on the this 
> platform (CloudSQL being as fragile as MySQL in that regard, and the 
> problem being compounded for me by the lack of tickets and my 
> uncertainty as to the fix given that the pickles are in the problem db). 
>
> I think perhaps that I too need to take an "old school and pessimistic" 
> approach on this platform (and MySQL in general for that matter). 
>
> BTW, in case anyone wonders why I'm using "fake_migrate_all" rather than 
> a per-table approach, it's because the error I was responding to 
> refereed to my (unmodified) auth_user table - I've simply not been able 
> to get in to revert that since this happened. 
>
>
> [1] https://code.google.com/p/googleappengine/issues/detail?id=8918 
>
> -- 
>
> Regards, 
>
> PhilK 
>
>
> e: ph...@xfr.co.uk  - m: 07775 796 747 
>
> 'work as if you lived in the early days of a better nation' 
> - alasdair gray 
>

-- 

--- 
You received this message because you are subscribed to the Google Groups 
"web2py-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to web2py+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.




[web2py] Re: GAE Hosting Issue

2013-03-04 Thread Philip Kilner

Hi Christian,

On 05/03/13 06:12, Christian Foster Howes wrote:

hrmmaybe it's time for GAE's next code update.  i frequently notice
a degradation in performance just before they announce a new SDK version.



Interesting observation - will look out for that.


i'm not seeing anything out of the ordinary on my GAE apps right now,
but i do agree that the GAE team is not as transparent as i would like.

i would suggest that you read the logs very carefully - there are lots
of different deadlineexceeded exceptions (literally the same class names
with different paths) and try adding extra logging to see where things
are getting stuck.



Actually, I'm not getting past a db connection so there is nothing in my 
application code to log, and the logged errors accessible in the console 
are very, very varied [1].



i don't know if web2py does automatic migrations on cloudsql, and in
production i always do my SQL by handbut i'm old school and pessimistic.



It does - I did have a failed migration (how I wish their CloudSQL 
offering was based on Postgres!), but the app was buzzing away happily 
with the help of "fake_migrate_all" for a good while after that.


The migration issues is messier than I'd like it to be on the this 
platform (CloudSQL being as fragile as MySQL in that regard, and the 
problem being compounded for me by the lack of tickets and my 
uncertainty as to the fix given that the pickles are in the problem db).


I think perhaps that I too need to take an "old school and pessimistic" 
approach on this platform (and MySQL in general for that matter).


BTW, in case anyone wonders why I'm using "fake_migrate_all" rather than 
a per-table approach, it's because the error I was responding to 
refereed to my (unmodified) auth_user table - I've simply not been able 
to get in to revert that since this happened.



[1] https://code.google.com/p/googleappengine/issues/detail?id=8918

--

Regards,

PhilK


e: p...@xfr.co.uk - m: 07775 796 747

'work as if you lived in the early days of a better nation'
- alasdair gray

--

--- 
You received this message because you are subscribed to the Google Groups "web2py-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email 
to web2py+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.




[web2py] Re: GAE Hosting Issue

2013-03-04 Thread Christian Foster Howes
hrmmaybe it's time for GAE's next code update.  i frequently notice a 
degradation in performance just before they announce a new SDK version.

i'm not seeing anything out of the ordinary on my GAE apps right now, but i 
do agree that the GAE team is not as transparent as i would like.

i would suggest that you read the logs very carefully - there are lots of 
different deadlineexceeded exceptions (literally the same class names with 
different paths) and try adding extra logging to see where things are 
getting stuck.

i don't know if web2py does automatic migrations on cloudsql, and in 
production i always do my SQL by handbut i'm old school and pessimistic.

cfh

On Monday, March 4, 2013 10:00:02 PM UTC-8, Philip Kilner wrote:
>
> Hi, 
>
> On 05/03/13 02:05, howesc wrote: 
> > i'm no longer using cloudSQL as i don't need it anymorebut several 
> > months ago when i was using it i got weird and varied problems on GAE 
> > when i had exceeded my billing quota on could SQL.  have you checked 
> > that those SQL instances are running and billing up to date? 
> > 
>
> I do have billing enabled, but am currently on a (temporarily) free D0 
> instance. Interesting that the problem kicked in less than an hour after 
> the start of the month though... 
>
> Having said that, I'm able to access the db perfectly via dbVisualiser 
> to work with it directly. 
>
> The majority of the errors suggest it is running out of time or 
> resources before it ever makes a db connection. 
>
> However, tinkering with a 2nd (empty) db against the same app it did 
> briefly burst into life, which made me wonder whether a failed migration 
> earlier in the day was involved. I can't see it really, because it never 
> seems to get as far as a connection with the production db. 
>
> This is in part the reason for my other question about using the 
> datastore form web2py's "system" stuff - I found myself wondering how 
> e.g. "fake_migrate_all" could work if the pickles were in the affected db. 
>
> I have to say, Google seem to be in denial about some of the issues on 
> GAE - some people (mostly but not exclusively on Java) have recently 
> suffered non-trivia performance / latency / timeout issues, and the 
> silence from the App Engine team is deafening... 
>
>
> -- 
>
> Regards, 
>
> PhilK 
>
>
> e: ph...@xfr.co.uk  - m: 07775 796 747 
>
> 'work as if you lived in the early days of a better nation' 
> - alasdair gray 
>

-- 

--- 
You received this message because you are subscribed to the Google Groups 
"web2py-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to web2py+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.




[web2py] Re: GAE Hosting Issue

2013-03-04 Thread Philip Kilner

Hi,

On 05/03/13 02:05, howesc wrote:

i'm no longer using cloudSQL as i don't need it anymorebut several
months ago when i was using it i got weird and varied problems on GAE
when i had exceeded my billing quota on could SQL.  have you checked
that those SQL instances are running and billing up to date?



I do have billing enabled, but am currently on a (temporarily) free D0 
instance. Interesting that the problem kicked in less than an hour after 
the start of the month though...


Having said that, I'm able to access the db perfectly via dbVisualiser 
to work with it directly.


The majority of the errors suggest it is running out of time or 
resources before it ever makes a db connection.


However, tinkering with a 2nd (empty) db against the same app it did 
briefly burst into life, which made me wonder whether a failed migration 
earlier in the day was involved. I can't see it really, because it never 
seems to get as far as a connection with the production db.


This is in part the reason for my other question about using the 
datastore form web2py's "system" stuff - I found myself wondering how 
e.g. "fake_migrate_all" could work if the pickles were in the affected db.


I have to say, Google seem to be in denial about some of the issues on 
GAE - some people (mostly but not exclusively on Java) have recently 
suffered non-trivia performance / latency / timeout issues, and the 
silence from the App Engine team is deafening...



--

Regards,

PhilK


e: p...@xfr.co.uk - m: 07775 796 747

'work as if you lived in the early days of a better nation'
- alasdair gray

--

--- 
You received this message because you are subscribed to the Google Groups "web2py-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email 
to web2py+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.




[web2py] Re: GAE Hosting Issue

2013-03-04 Thread howesc
i'm no longer using cloudSQL as i don't need it anymorebut several 
months ago when i was using it i got weird and varied problems on GAE when 
i had exceeded my billing quota on could SQL.  have you checked that those 
SQL instances are running and billing up to date?

On Monday, March 4, 2013 3:31:22 AM UTC-8, Philip Kilner wrote:
>
> Hi All, 
>
> I'm developing an app on GAE which has suddenly stopped serving, and 
> just wanted to sanity check that other web2py users were not seeing the 
> same thing. 
>
> The app uses Python 2.7, HRD and CloudSQL. It's a relatively simple 
> transactional CRUD app - nothing clever at all from the GAE PoV. 
>
> It stopped serving out of the blue (overnight), twelve hours after the 
> last deploy, and the error messages it's throwing are very varied - 
> "soft memory limit exceeded" (bumped instances from F1 to F2, same 
> error, bigger numbers), "deadline exceeded" (setting up loggers or db 
> connections) etc. never get as far as my code. 
>
> The lack of changes in the app prior to the outage, the essentially 
> random nature of the error messages (all of which suggest GAE is running 
> out of time or resources in a variety of interesting ways), suggest to 
> me that this is a GAE infrastructure problem, but I've yet to find 
> anyone else definitively bitten by the same thing (although there is a 
> lot of grumbling in the GAE forums about high latency). 
>
> I've raised a production issue [1], so far with no response, and have 
> moved the app off GAE so that I can carry on working, so the pressure is 
> off. 
>
> Has anyone else using this combination of GAE facilities (and especially 
> CloudSQL) experienced anything similar? 
>
> FWIW, my (relatively trivial) apps running purely on the normal HRD 
> data-store are running fine. 
>
>
> [1] https://code.google.com/p/googleappengine/issues/detail?id=8918 
>
> -- 
>
> Regards, 
>
> PhilK 
>
>
> e: ph...@xfr.co.uk  - m: 07775 796 747 
>
> 'work as if you lived in the early days of a better nation' 
> - alasdair gray 
>

-- 

--- 
You received this message because you are subscribed to the Google Groups 
"web2py-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to web2py+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.