Double LMDB map_size on MapFullError #209
Conversation
| curr_limit = self.db.info()['map_size'] | ||
| new_limit = curr_limit*2 | ||
| logger.debug('Doubling LMDB map size to %sMB ...' % (new_limit>>20,)) | ||
| self.db.set_mapsize(curr_limit*2) # double it |
There was a problem hiding this comment.
since this is a new API in py-lmdb, is it worth adding a check that version is >0.86? I notice the version has not been updated yet in https://github.com/dw/py-lmdb/blob/master/lmdb/__init__.py but I guess this will come soon.
There was a problem hiding this comment.
Hi @gheinrich, I see the problem now. Somehow I committed the version bump to the release branch, not master. I've merged the single missing commit back to master. Thanks for catching!
There was a problem hiding this comment.
Just to be clear, the release branch and all released wheels, tarballs, git tags and GitHub tarballs will include the correct __init__.py, the problem only affected master branch.
lukeyeager
force-pushed
the
lmdb-map-size
branch
from
471b37e
to
0528d1e
Compare
|
Ok, I've checked off all the TODOs. @crohkohl, could you run the test suite on Windows when you get a chance? |
|
This looks good to me. |
|
@lukeyeager Windows tests finally pass, see #199 (comment) |
lukeyeager
force-pushed
the
lmdb-map-size
branch
from
2c654ab
to
c9371ce
Compare
Now requires py-lmdb >= 0.87
lukeyeager
force-pushed
the
lmdb-map-size
branch
from
c9371ce
to
944bea6
Compare
|
Squashed to a single commit and rebased to include TravisCI fix in #214. |
Double LMDB map_size on MapFullError
|
@gheinrich and @crohkohl, thanks for the review help. |
Closes #206.
Changed
TODO before merge
digits/dataset/images/generic/test_lmdb_creator.pyas well