We managed to get a fun stacktrace on a node that was simultaneously
generating the bytecode cache and another process that was trying to read
it.
Looking at class FileSystemBytecodeCache in bccache.py, I think this
would've been avoided if the dump_bytecode function wrote to a tmp file and
then did an atomic rename, e.g. something like:
def dump_bytecode(self, bucket):
_cache_file = self._get_cache_filename(bucket)
_cache_tmp_file = os.path.join(_cache_file, '-tmp-%s' % os.getpid())
f = open(_cache_tmp_file, 'wb')
try:
bucket.write_bytecode(f)
os.rename(_cache_tmp_file, _cache_file)
finally:
f.close()
Thoughts?
-Jeff
--
You received this message because you are subscribed to the Google Groups
"pocoo-libs" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/pocoo-libs.
For more options, visit https://groups.google.com/groups/opt_out.
