continual transient error on pull queue

[c h]

Hello,

I have a pull queue that i have been getting:

  File "/base/python_runtime/python_lib/versions/1/google/appengine/api/taskqueue/taskqueue.py", line 1423, in lease_tasks
    raise self.__TranslateError(e.application_error, e.error_detail)
TransientError

for the past day.  to put it in perspective - i had a problem with the processing of tasks and shut off my consumer of this queue for a few days to fix some stuff, i turned the consumer back on yesterday and have not been able to read a task since.  is this because my pull queue has gotten quite large (over 6gb, and the count is unable to be displayed in the console, probably several million items)?  is there anything i can do to recover, or do i just need to purge the queue and start over?  (given the data that is in the queue i would much prefer to recover)

thanks,

christian

[Nicholas Verne]

Christian,

Your problem sounds like an instance of a rare condition in our
backends that can cause deadlock in a pull queue. We are working on a
fix that should be pushed out in the next few days. If you're prepared
to wait a couple more days, you may be able to consume your remaining
tasks without needing to purge the queue. Until then, I advise against
adding new tasks to the queue.

Nick Verne

> --
> You received this message because you are subscribed to the Google Groups
> "Google App Engine" group.
> To view this discussion on the web visit
> https://groups.google.com/d/msg/google-appengine/-/NGrhdc9xAkIJ.
> To post to this group, send email to google-a...@googlegroups.com.
> To unsubscribe from this group, send email to
> google-appengi...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/google-appengine?hl=en.

[Christian Foster Howes]

Nick,

thanks for the info! 2 follow-on questions:
- do you expect that i can read the queue from a front-end instance?
- where do i sign-up to be notified that a fix is in place? (is there
a bug number i can follow?)

thanks so much for the help, and the great product!

christian

[Nicholas Verne]

If the queue is deadlocked in the way I suspect, it will remain so
regardless of whether your app accesses it from a backend or frontend
instance.

I'll respond in this thread.

Nick Verne

[Christian Foster Howes]

thank you!

i'll stop writing to it as soon as i can get the change in my code. :)

please let me know if you need help from me to debug or test.

christian

[Nicholas Verne]

Christian,

The presumed fix was pushed out to production a few hours ago. If you
want to try leasing from your eventpull queue again, you should have a
better rate of success.

Nick Verne

On Tue, Feb 28, 2012 at 10:45 AM, Christian Foster Howes

[c h]

Nick,

Thanks much!  i started leasing tasks again and it appears to be working.  I've been through a few thousand tasks, only 19 million more tasks to go!

thanks for keeping me updated.

christian