diff options
| -rw-r--r-- | python/README_import.md | 18 |
1 files changed, 16 insertions, 2 deletions
diff --git a/python/README_import.md b/python/README_import.md index 7301d72e..f43d9424 100644 --- a/python/README_import.md +++ b/python/README_import.md @@ -23,8 +23,13 @@ the others: From CSV file: + export LC_ALL=C.UTF-8 time ./client.py import-issn /srv/datasets/journal_extra_metadata.csv + real 2m42.148s + user 0m11.148s + sys 0m0.336s + Pretty quick, a few minutes. ## ORCID @@ -33,7 +38,8 @@ Directly from compressed tarball; takes about 2 hours in production: tar xf /srv/datasets/public_profiles_API-2.0_2017_10_json.tar.gz -O | jq -c . | grep '"person":' | time parallel -j12 --pipe --round-robin ./client.py import-orcid - -Or, from pre-uncompressed tarball: +After tuning database, `jq` CPU seems to be bottleneck, so, from pre-extracted +tarball: tar xf /srv/datasets/public_profiles_API-2.0_2017_10_json.tar.gz -O | jq -c . | rg '"person":' > /srv/datasets/public_profiles_1_2_json.all.json time parallel --bar --pipepart -j8 -a /srv/datasets/public_profiles_1_2_json.all.json ./client.py import-orcid - @@ -76,11 +82,19 @@ For the full batch, on production machine with 12 threads, around 3.8 million re => 644/second +After some simple database tuning: + + 2177.86 user + 145.60 system + 56:41.26 elapsed + + => 1117/second + ## Crossref From compressed: - xzcat /srv/datasets/crossref-works.2018-01-21.json.xz | time parallel -j12 --round-robin --pipe ./client.py import-crossref - /srv/datasets/20180216.ISSN-to-ISSN-L.txt + xzcat /srv/datasets/crossref-works.2018-01-21.json.xz | time parallel -j20 --round-robin --pipe ./client.py import-crossref - /srv/datasets/20180216.ISSN-to-ISSN-L.txt ## Manifest |