swh.indexer.metadata should not use directory_ls
The swh.indexer.metadata
module uses self.storage.directory_ls()
in order to get all entries in a directory.
However, there may be many entries, causing the server to crash because of timeouts (which is why this method is deprecated). Even if it worked, it could take a significant amount of memory.
Instead, swh.indexer.metadata
should use self.storage.directory_get_entries()
, which supports pagination.
Migrated from T4269 (view on Phabricator)
Edited by Phabricator Migration user