diff options
author | Bryan Newbold <bnewbold@archive.org> | 2021-04-09 13:03:18 -0700 |
---|---|---|
committer | Bryan Newbold <bnewbold@archive.org> | 2021-04-09 13:03:18 -0700 |
commit | dd4e8bcd6dbe0aaa141b94c851a47963bb49563a (patch) | |
tree | 72ab1676a12683caf63cd1676b1592f98fa5ec28 /sql/reingest_quarterly.sh | |
parent | e40d7ddf0c091426e8c61cc1a5be5f15bd052c10 (diff) | |
download | sandcrawler-dd4e8bcd6dbe0aaa141b94c851a47963bb49563a.tar.gz sandcrawler-dd4e8bcd6dbe0aaa141b94c851a47963bb49563a.zip |
sql: update periodic retry/reingest scripts
Diffstat (limited to 'sql/reingest_quarterly.sh')
-rwxr-xr-x | sql/reingest_quarterly.sh | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/sql/reingest_quarterly.sh b/sql/reingest_quarterly.sh index 44a22b3..8b0889d 100755 --- a/sql/reingest_quarterly.sh +++ b/sql/reingest_quarterly.sh @@ -9,12 +9,12 @@ sudo -u postgres psql sandcrawler < dump_reingest_quarterly.sql cd ../python pipenv run ./scripts/ingestrequest_row2json.py /grande/snapshots/reingest_quarterly_spn2-error_current.rows.json | shuf > /grande/snapshots/reingest_quarterly_spn2-error_current.json pipenv run ./scripts/ingestrequest_row2json.py /grande/snapshots/reingest_quarterly_cdx-error_current.rows.json | shuf > /grande/snapshots/reingest_quarterly_cdx-error_current.json -pipenv run ./scripts/ingestrequest_row2json.py /grande/snapshots/reingest_quarterly_cdx-error_bulk_current.rows.json | shuf > /grande/snapshots/reingest_quarterly_cdx-error_bulk_current.json +#pipenv run ./scripts/ingestrequest_row2json.py /grande/snapshots/reingest_quarterly_cdx-error_bulk_current.rows.json | shuf > /grande/snapshots/reingest_quarterly_cdx-error_bulk_current.json pipenv run ./scripts/ingestrequest_row2json.py /grande/snapshots/reingest_quarterly_wayback-error_current.rows.json | shuf > /grande/snapshots/reingest_quarterly_wayback-error_current.json pipenv run ./scripts/ingestrequest_row2json.py /grande/snapshots/reingest_quarterly_gateway-timeout.rows.json | shuf > /grande/snapshots/reingest_quarterly_gateway-timeout.json pipenv run ./scripts/ingestrequest_row2json.py /grande/snapshots/reingest_quarterly_petabox-error_current.rows.json | shuf > /grande/snapshots/reingest_quarterly_petabox-error_current.json -cat /grande/snapshots/reingest_quarterly_spn2-error_current.json /grande/snapshots/reingest_quarterly_cdx-error_current.json /grande/snapshots/reingest_quarterly_wayback-error_current.json /grande/snapshots/reingest_quarterly_petabox-error_current.json /grande/snapshots/reingest_quarterly_gateway-timeout.json | shuf | head -n100000 | jq . -c | kafkacat -P -b wbgrp-svc263.us.archive.org -t sandcrawler-prod.ingest-file-requests -p -1 +cat /grande/snapshots/reingest_quarterly_spn2-error_current.json /grande/snapshots/reingest_quarterly_cdx-error_current.json /grande/snapshots/reingest_quarterly_wayback-error_current.json /grande/snapshots/reingest_quarterly_petabox-error_current.json /grande/snapshots/reingest_quarterly_gateway-timeout.json | shuf | head -n250000 | jq . -c | kafkacat -P -b wbgrp-svc263.us.archive.org -t sandcrawler-prod.ingest-file-requests -p -1 -cat /grande/snapshots/reingest_quarterly_cdx-error_bulk.json | shuf | jq . -c | kafkacat -P -b wbgrp-svc263.us.archive.org -t sandcrawler-prod.ingest-file-requests-bulk -p -1 +#cat /grande/snapshots/reingest_quarterly_cdx-error_bulk.json | shuf | jq . -c | kafkacat -P -b wbgrp-svc263.us.archive.org -t sandcrawler-prod.ingest-file-requests-bulk -p -1 |