IndexError in TableMapEvent

[lxyu]

Traceback (most recent call last):
  File "/srv/virtualenvs/zeus/local/lib/python2.7/site-packages/pymysqlreplication/binlogstream.py", line 118, in fetchone
    self._ctl_connection)
  File "/srv/virtualenvs/zeus/local/lib/python2.7/site-packages/pymysqlreplication/packet.py", line 88, in __init__
    ctl_connection)
  File "/srv/virtualenvs/zeus/local/lib/python2.7/site-packages/pymysqlreplication/row_event.py", line 488, in __init__
    column_schema = self.column_schemas[i]
IndexError: tuple index out of range

More info:

>>> print column_types
['\x03', '\x03', '\x03', '\x0f', '\x0f', '\x03', '\x0c', '\x0c']
>>> print self.columns
[]
>>> print table_map
{}
>>> print self.column_schemas
()

And the self.packet.read_length_coded_binary() at L485 returns 4.

Put this line in the first line of TableMapEvent init shows this:

>>> print repr(self.packet.read(58))
'N\x01\x00\x00\x00\x00\x01\x00\x06sentry\x00\x19sentry_messagefiltervalue\x00\x08\x03\x03\x03\x0f\x0f\x03\x0c\x0c\x04`\x00X\x02\xe0'

[dvirsky]

what caused this to happen? a rotation maybe?

[lxyu]

I don't know what caused it, knows nothing more than the packet info pasted above.

It happened in our production server, very rare but it happened randomly. And I just bypass the table map event cause I don't really care about it.

[bjoernhaeuser]

The problem is probably that you try to create events for a table which is already dropped. Just figured out that I have the same problem. The application creates a table and after finishing the work it gets dropped. This module tries afterwards to decode the table data and fails, because it cannot read the information out of the information_schema.

I am currently thinking of a fix but haven't figured out a clean way, maybe @noplay can jump in

[bjoernhaeuser]

This is in https://github.com/noplay/python-mysql-replication/tree/indexerror

[julien-duponchelle]

You are right i'm not a big fan of your fix, but i didn't see a good way to do that.

[lxyu]

@bjoernhaeuser my tmp fix is very like your fix.

refer to lxyu@a384964

I only add one line to skip it, though.

[bjoernhaeuser]

@noplay can we can up with a more elegant solution? What are your thoughts?

[julien-duponchelle]

No idea about a better way :( Perhaps we can merge and add an issue about that as a reminder.

[bjoernhaeuser]

I will come up with a pull request today.

[psypuff]

Hi, looks like this issue has been open for a while... 😁

We've recently got this similar exception on our production env:

Traceback (most recent call last):
  File "my_script.py", line 107, in <module>
    for binlogevent in stream:
  File "/usr/local/lib/python2.7/dist-packages/pymysqlreplication/binlogstream.py", line 244, in fetchone
    only_schemas = self.__only_schemas)
  File "/usr/local/lib/python2.7/dist-packages/pymysqlreplication/packet.py", line 94, in __init__
    only_schemas = only_schemas)
  File "/usr/local/lib/python2.7/dist-packages/pymysqlreplication/row_event.py", line 512, in __init__
    column_schema = self.column_schemas[i]
IndexError: tuple index out of range

I guess the issue is still open `cause there's no fix yet...
Is there any fix planned?

[julien-duponchelle]

@bjoernhaeuser @lxyu what did you think? I can add the fix you proposed if you didn't have another idea since.

[bjoernhaeuser]

@noplay No idea from me :-(

[julien-duponchelle]

0.6 has been released with the fix from @bjoernhaeuser