So I'm working to scale up my web2py based app a bit. Part of this was
moving the Scheduler to a separate machine. Since doing this I'm getting
some errors and weird behavior and could use some insight. There are
multiple issues but I sort of need to explain it all so I opted for 1 post
with multiple questions instead of multiple concise posts...sorry
My configuration:
"production server" - nginx serving my website. db connects to a mysql
instance bound to production server network address. scheduler connects to
mysql instance running on the "dev/workq server"
"dev/workq server" - nginx serving a copy of the same web2py
directory...plan to use as development server if needed as well. db
connects to mysql instance running on production server. scheduler connects
to mysql instance bound to dev server network address.
from 0_db.py:
db =
DAL('mysql://dev:xxx...@production.server.edu/myapp',pool_size=8,check_reserved=['mysql'],migrate=ENABLE_MIGRATE,fake_migrate_all=ENABLE_FAKE_MIGRATE)
from scheduler.py:
scheduler = Scheduler(DAL('mysql://workq:x...@dev.workq.server.edu/myapp
',pool_size=8,check_reserved=['mysql'],migrate=ENABLE_MIGRATE,fake_migrate_all=ENABLE_FAKE_MIGRATE),heartbeat=2)
My steps from empty mysql databases and database directories on both
servers and both mysql instances.
- go to site on the production server - migrate = True, fake_migrate =
False -> OK
- go to site on the dev server - migrate = True, fake_migrate = False ->
Error - <class 'gluon.contrib.pymysql.err.InternalError'> (1050, u"Table
'auth_user' already exists")
- go to site on the dev server - migrate = False, fake_migrate = True -> OK
- start scheduler task -> OK
*Am I starting this all up improperly? I'm a little confused since I've
got two web2py instances talking to different db instances for the web app
and the scheduler...but I think having to do a fake migrate on the second
server makes sense.*
*
*
*
*
So now I *think* my website is up and running properly. I then run a
function that schedules a job. This seems to run (by looking at the
comfortscheduler monitor), but it's supposed to schedule additional jobs
itself. This never happens.
inside the single task that is scheduled and runs:
for items in my_thing:
#...do stuff
# Submit tasks to process more stuff
print "submitting job \n"
scheduler.queue_task(my_task2,timeout=60000,
pvars=dict(arg1= arg1,
arg2=arg2))
db(db.collections.name == collection).update(last_id=last_id)
db.commit()
*If I try to view the details of the task in the ComfortScheduler monitor
(from my production server) by clicking on the UUID link of the task i get
an error:*
<type 'exceptions.AttributeError'> 'DAL' object has no attribute
'scheduler_task'
1.
2.
3.
4.
5.
6.
7.
8.
Traceback (most recent call last):
File "/home/www-data/web2py/gluon/restricted.py", line 217, in restricted
exec ccode in environment
File
"/home/www-data/web2py/applications/parity/views/plugin_cs_monitor/task_details.html",
line 111, in <module>
File "/home/www-data/web2py/gluon/dal.py", line 8041, in __getattr__
return ogetattr(self, key)
AttributeError: 'DAL' object has no attribute 'scheduler_task'
I think the problem here is the comfortscheduler code looks at the global
db object maybe? Since I've passed in to Scheduler() a different database
is this breaking things? This might be a question only niphlod can answer
since it's his app....
So ignoring that for now, if i go and look in the table I can see the
run_output "submitting job submitting job submitting job submitting job"
indicating
it got to the point in the code where it should have submitted more tasks.
*Any idea why new tasks would not be getting scheduled? *I think it might
be because I'm calling db.commit() in my task...but this is on my main
web2py db, not the db the scheduler is using? Can i have 2 global db
objects? So should the scheduler be setup more like this:
sched_db = DAL('mysql://workq:x...@dev.workq.server.edu/myapp
',pool_size=8,check_reserved=['mysql'],migrate=ENABLE_MIGRATE,fake_migrate_all=ENABLE_FAKE_MIGRATE)
scheduler = Scheduler(sched_db,heartbeat=2)
and then i'd have to call both
db.commit()
sched_db.commit()
Anyone have a server config like this before?
Long message...lots of questions...sorry and thanks.
Dean
--
Resources:
- http://web2py.com
- http://web2py.com/book (Documentation)
- http://github.com/web2py/web2py (Source code)
- https://code.google.com/p/web2py/issues/list (Report Issues)
---
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.
