summaryrefslogtreecommitdiffstats
path: root/extra
diff options
context:
space:
mode:
Diffstat (limited to 'extra')
-rw-r--r--extra/docker/Dockerfile.test-base47
-rw-r--r--extra/docker/README.md4
-rw-r--r--extra/stats/2020-02-21-prod-stats.json1
-rw-r--r--extra/stats/2020-02-21-prod-tables-sizes.txt48
-rw-r--r--extra/stats/2020-02-24-prod-dupes.txt5
-rw-r--r--extra/stats/2020-02-24-prod-table-sizes.txt47
-rw-r--r--extra/stats/2020-03-03-prod-stats.json1
-rw-r--r--extra/stats/2020-04-17-prod-stats.json1
-rw-r--r--extra/stats/2020-04-17-prod-table-sizes.txt46
9 files changed, 200 insertions, 0 deletions
diff --git a/extra/docker/Dockerfile.test-base b/extra/docker/Dockerfile.test-base
new file mode 100644
index 00000000..a556ed99
--- /dev/null
+++ b/extra/docker/Dockerfile.test-base
@@ -0,0 +1,47 @@
+
+FROM ubuntu:xenial
+
+ENV RUSTUP_HOME=/usr/local/rustup \
+ CARGO_HOME=/usr/local/cargo \
+ PATH=/usr/local/cargo/bin:$PATH \
+ RUST_VERSION=1.42.0 \
+ LC_ALL=C.UTF-8 \
+ LANG=C.UTF-8
+
+
+# Add deadsnakes repo
+RUN set -eux; \
+ apt update -qy; \
+ apt install -y software-properties-common; \
+ add-apt-repository -y ppa:deadsnakes/ppa;
+
+# APT dependencies
+RUN set -eux; \
+ apt update -qy; \
+ apt install -y python3-dev python3-pip python3-wheel python3-requests python3-six python3-pytest libsnappy-dev libsodium-dev software-properties-common python3.7 python3.7-dev python3.7-venv python3.7-distutils wget libpq-dev pkg-config python3-pytest git
+
+# Rust setup from docker-rust debian Dockerfile
+RUN set -eux; \
+ dpkgArch="$(dpkg --print-architecture)"; \
+ rustArch='x86_64-unknown-linux-gnu'; rustupSha256='ad1f8b5199b3b9e231472ed7aa08d2e5d1d539198a15c5b1e53c746aad81d27b' ; \
+ url="https://static.rust-lang.org/rustup/archive/1.21.1/${rustArch}/rustup-init"; \
+ wget "$url"; \
+ echo "${rustupSha256} *rustup-init" | sha256sum -c -; \
+ chmod +x rustup-init; \
+ ./rustup-init -y --no-modify-path --profile minimal --default-toolchain $RUST_VERSION; \
+ rm rustup-init; \
+ chmod -R a+w $RUSTUP_HOME $CARGO_HOME; \
+ rustup --version; \
+ cargo --version; \
+ rustc --version;
+
+# Compile and install diesel
+RUN set -eux; \
+ cargo install diesel_cli --version 1.3.1 --no-default-features --features postgres; \
+ diesel --version
+
+# Install pipenv
+RUN set -eux; \
+ pip3 install pipenv; \
+ pipenv --version
+
diff --git a/extra/docker/README.md b/extra/docker/README.md
index 4fd78e56..1869b354 100644
--- a/extra/docker/README.md
+++ b/extra/docker/README.md
@@ -17,3 +17,7 @@ TODO:
- postgres
- fatcatd (rust)
- kibana
+
+## Test Base Image
+
+ docker build -t bnewbold/fatcat-test-base -f Dockerfile.test-base .
diff --git a/extra/stats/2020-02-21-prod-stats.json b/extra/stats/2020-02-21-prod-stats.json
new file mode 100644
index 00000000..3ab6471f
--- /dev/null
+++ b/extra/stats/2020-02-21-prod-stats.json
@@ -0,0 +1 @@
+{"changelog":{"latest":{"index":3528195,"timestamp":"2020-02-22T05:23:18.082262+00:00"}},"container":{"total":148396},"papers":{"in_kbart":60529767,"in_web":20374670,"in_web_not_kbart":9598464,"is_oa":11547112,"total":105732384},"release":{"refs_total":890869519,"total":143867045}}
diff --git a/extra/stats/2020-02-21-prod-tables-sizes.txt b/extra/stats/2020-02-21-prod-tables-sizes.txt
new file mode 100644
index 00000000..bc756ba7
--- /dev/null
+++ b/extra/stats/2020-02-21-prod-tables-sizes.txt
@@ -0,0 +1,48 @@
+
+Size: 478.37G
+
+ table_name | table_size | indexes_size | total_size
+---------------------------------------+------------+--------------+------------
+ "public"."release_contrib" | 53 GB | 43 GB | 96 GB
+ "public"."release_rev" | 58 GB | 33 GB | 91 GB
+ "public"."refs_blob" | 85 GB | 2884 MB | 88 GB
+ "public"."release_edit" | 14 GB | 20 GB | 34 GB
+ "public"."work_edit" | 13 GB | 20 GB | 34 GB
+ "public"."release_ident" | 9515 MB | 15 GB | 24 GB
+ "public"."work_ident" | 9313 MB | 15 GB | 24 GB
+ "public"."abstracts" | 16 GB | 1504 MB | 18 GB
+ "public"."file_rev_url" | 10235 MB | 3587 MB | 13 GB
+ "public"."work_rev" | 6046 MB | 5825 MB | 12 GB
+ "public"."release_ref" | 3997 MB | 5690 MB | 9686 MB
+ "public"."file_rev" | 3635 MB | 5359 MB | 8994 MB
+ "public"."file_edit" | 3111 MB | 4051 MB | 7162 MB
+ "public"."release_rev_abstract" | 2406 MB | 3342 MB | 5749 MB
+ "public"."file_ident" | 1848 MB | 2505 MB | 4354 MB
+ "public"."file_rev_release" | 1698 MB | 2483 MB | 4181 MB
+ "public"."creator_edit" | 702 MB | 942 MB | 1643 MB
+ "public"."creator_rev" | 695 MB | 719 MB | 1413 MB
+ "public"."editgroup" | 767 MB | 405 MB | 1172 MB
+ "public"."creator_ident" | 474 MB | 648 MB | 1121 MB
+ "public"."release_rev_extid" | 200 MB | 312 MB | 512 MB
+ "public"."changelog" | 220 MB | 214 MB | 434 MB
+ "public"."container_rev" | 75 MB | 23 MB | 98 MB
+ "public"."container_edit" | 25 MB | 31 MB | 56 MB
+ "public"."container_ident" | 11 MB | 19 MB | 30 MB
+ "public"."webcapture_rev_cdx" | 64 kB | 32 kB | 96 kB
+ "public"."fileset_rev_file" | 48 kB | 32 kB | 80 kB
+ "public"."auth_oidc" | 16 kB | 48 kB | 64 kB
+ "public"."fileset_edit" | 16 kB | 48 kB | 64 kB
+ "public"."editor" | 16 kB | 48 kB | 64 kB
+ "public"."webcapture_edit" | 16 kB | 48 kB | 64 kB
+ "public"."editgroup_annotation" | 16 kB | 48 kB | 64 kB
+ "public"."fileset_rev_url" | 16 kB | 32 kB | 48 kB
+ "public"."webcapture_rev_url" | 16 kB | 32 kB | 48 kB
+ "public"."fileset_rev_release" | 8192 bytes | 32 kB | 40 kB
+ "public"."fileset_ident" | 8192 bytes | 32 kB | 40 kB
+ "public"."webcapture_rev_release" | 8192 bytes | 32 kB | 40 kB
+ "public"."webcapture_ident" | 8192 bytes | 32 kB | 40 kB
+ "public"."fileset_rev" | 16 kB | 16 kB | 32 kB
+ "public"."webcapture_rev" | 16 kB | 16 kB | 32 kB
+ "public"."__diesel_schema_migrations" | 8192 bytes | 16 kB | 24 kB
+(41 rows)
+
diff --git a/extra/stats/2020-02-24-prod-dupes.txt b/extra/stats/2020-02-24-prod-dupes.txt
new file mode 100644
index 00000000..7d1d09cf
--- /dev/null
+++ b/extra/stats/2020-02-24-prod-dupes.txt
@@ -0,0 +1,5 @@
+ 19409 doi_ident.dupes.tsv
+ 28530 pmcid_ident.dupes.tsv
+ 463523 pmid_ident.dupes.tsv
+ 2025 sha1_ident.dupes.tsv
+ 10 wikidata_ident.dupes.tsv
diff --git a/extra/stats/2020-02-24-prod-table-sizes.txt b/extra/stats/2020-02-24-prod-table-sizes.txt
new file mode 100644
index 00000000..359cb2f3
--- /dev/null
+++ b/extra/stats/2020-02-24-prod-table-sizes.txt
@@ -0,0 +1,47 @@
+
+Size: 560.76G
+
+ table_name | table_size | indexes_size | total_size
+---------------------------------------+------------+--------------+------------
+ "public"."release_contrib" | 53 GB | 43 GB | 96 GB
+ "public"."release_rev" | 58 GB | 33 GB | 91 GB
+ "public"."refs_blob" | 85 GB | 2884 MB | 88 GB
+ "public"."file_rev" | 23 GB | 26 GB | 49 GB
+ "public"."release_edit" | 14 GB | 20 GB | 34 GB
+ "public"."work_edit" | 13 GB | 20 GB | 34 GB
+ "public"."release_ident" | 9517 MB | 15 GB | 24 GB
+ "public"."work_ident" | 9315 MB | 15 GB | 24 GB
+ "public"."file_edit" | 9555 MB | 14 GB | 24 GB
+ "public"."abstracts" | 16 GB | 1505 MB | 18 GB
+ "public"."file_rev_url" | 13 GB | 4730 MB | 17 GB
+ "public"."file_ident" | 5885 MB | 9480 MB | 15 GB
+ "public"."file_rev_release" | 5515 MB | 9536 MB | 15 GB
+ "public"."work_rev" | 6047 MB | 5825 MB | 12 GB
+ "public"."release_ref" | 3997 MB | 5690 MB | 9686 MB
+ "public"."release_rev_abstract" | 2408 MB | 3343 MB | 5751 MB
+ "public"."creator_edit" | 702 MB | 942 MB | 1643 MB
+ "public"."creator_rev" | 695 MB | 719 MB | 1413 MB
+ "public"."editgroup" | 903 MB | 465 MB | 1368 MB
+ "public"."creator_ident" | 474 MB | 648 MB | 1121 MB
+ "public"."release_rev_extid" | 200 MB | 312 MB | 512 MB
+ "public"."changelog" | 261 MB | 229 MB | 490 MB
+ "public"."container_rev" | 75 MB | 23 MB | 98 MB
+ "public"."container_edit" | 25 MB | 31 MB | 56 MB
+ "public"."container_ident" | 11 MB | 19 MB | 30 MB
+ "public"."webcapture_rev_cdx" | 64 kB | 32 kB | 96 kB
+ "public"."fileset_rev_file" | 48 kB | 32 kB | 80 kB
+ "public"."auth_oidc" | 16 kB | 48 kB | 64 kB
+ "public"."editgroup_annotation" | 16 kB | 48 kB | 64 kB
+ "public"."fileset_edit" | 16 kB | 48 kB | 64 kB
+ "public"."webcapture_edit" | 16 kB | 48 kB | 64 kB
+ "public"."editor" | 16 kB | 48 kB | 64 kB
+ "public"."fileset_rev_url" | 16 kB | 32 kB | 48 kB
+ "public"."webcapture_rev_url" | 16 kB | 32 kB | 48 kB
+ "public"."fileset_rev_release" | 8192 bytes | 32 kB | 40 kB
+ "public"."fileset_ident" | 8192 bytes | 32 kB | 40 kB
+ "public"."webcapture_rev_release" | 8192 bytes | 32 kB | 40 kB
+ "public"."webcapture_ident" | 8192 bytes | 32 kB | 40 kB
+ "public"."fileset_rev" | 16 kB | 16 kB | 32 kB
+ "public"."webcapture_rev" | 16 kB | 16 kB | 32 kB
+ "public"."__diesel_schema_migrations" | 8192 bytes | 16 kB | 24 kB
+(41 rows)
diff --git a/extra/stats/2020-03-03-prod-stats.json b/extra/stats/2020-03-03-prod-stats.json
new file mode 100644
index 00000000..0ac977b8
--- /dev/null
+++ b/extra/stats/2020-03-03-prod-stats.json
@@ -0,0 +1 @@
+{"changelog":{"latest":{"index":4242658,"timestamp":"2020-03-03T18:35:06.153130+00:00"}},"container":{"total":148428},"papers":{"in_kbart":60594053,"in_web":22232097,"in_web_not_kbart":10756782,"is_oa":15267353,"total":105933568},"release":{"refs_total":893136234,"total":144138471}}
diff --git a/extra/stats/2020-04-17-prod-stats.json b/extra/stats/2020-04-17-prod-stats.json
new file mode 100644
index 00000000..ddf7fca1
--- /dev/null
+++ b/extra/stats/2020-04-17-prod-stats.json
@@ -0,0 +1 @@
+{"changelog":{"latest":{"index":4460684,"timestamp":"2020-04-17T18:03:34.373631+00:00"}},"container":{"total":149527},"papers":{"in_kbart":60679890,"in_web":24250766,"in_web_not_kbart":11970984,"is_oa":15538739,"total":108761510},"release":{"refs_total":914708032,"total":148081134}}
diff --git a/extra/stats/2020-04-17-prod-table-sizes.txt b/extra/stats/2020-04-17-prod-table-sizes.txt
new file mode 100644
index 00000000..79aa3b98
--- /dev/null
+++ b/extra/stats/2020-04-17-prod-table-sizes.txt
@@ -0,0 +1,46 @@
+Size: 591.60G
+
+ table_name | table_size | indexes_size | total_size
+---------------------------------------+------------+--------------+------------
+ "public"."release_contrib" | 55 GB | 45 GB | 100 GB
+ "public"."release_rev" | 60 GB | 34 GB | 94 GB
+ "public"."refs_blob" | 87 GB | 2885 MB | 89 GB
+ "public"."file_rev" | 26 GB | 29 GB | 55 GB
+ "public"."release_edit" | 14 GB | 21 GB | 35 GB
+ "public"."work_edit" | 14 GB | 21 GB | 34 GB
+ "public"."file_edit" | 11 GB | 16 GB | 27 GB
+ "public"."release_ident" | 9821 MB | 15 GB | 24 GB
+ "public"."work_ident" | 9596 MB | 15 GB | 24 GB
+ "public"."file_rev_url" | 15 GB | 6040 MB | 21 GB
+ "public"."abstracts" | 18 GB | 1688 MB | 19 GB
+ "public"."file_ident" | 6694 MB | 10219 MB | 17 GB
+ "public"."file_rev_release" | 6267 MB | 10109 MB | 16 GB
+ "public"."work_rev" | 6233 MB | 5825 MB | 12 GB
+ "public"."release_ref" | 4441 MB | 6322 MB | 11 GB
+ "public"."release_rev_abstract" | 2637 MB | 3505 MB | 6141 MB
+ "public"."creator_edit" | 702 MB | 942 MB | 1643 MB
+ "public"."editgroup" | 980 MB | 502 MB | 1482 MB
+ "public"."creator_rev" | 695 MB | 719 MB | 1413 MB
+ "public"."creator_ident" | 474 MB | 648 MB | 1121 MB
+ "public"."changelog" | 289 MB | 239 MB | 527 MB
+ "public"."release_rev_extid" | 206 MB | 320 MB | 526 MB
+ "public"."container_rev" | 75 MB | 23 MB | 98 MB
+ "public"."container_edit" | 25 MB | 32 MB | 57 MB
+ "public"."container_ident" | 11 MB | 19 MB | 30 MB
+ "public"."webcapture_rev_cdx" | 64 kB | 32 kB | 96 kB
+ "public"."fileset_rev_file" | 48 kB | 32 kB | 80 kB
+ "public"."auth_oidc" | 16 kB | 48 kB | 64 kB
+ "public"."editgroup_annotation" | 16 kB | 48 kB | 64 kB
+ "public"."fileset_edit" | 16 kB | 48 kB | 64 kB
+ "public"."webcapture_edit" | 16 kB | 48 kB | 64 kB
+ "public"."editor" | 16 kB | 48 kB | 64 kB
+ "public"."fileset_rev_url" | 16 kB | 32 kB | 48 kB
+ "public"."webcapture_rev_url" | 16 kB | 32 kB | 48 kB
+ "public"."fileset_rev_release" | 8192 bytes | 32 kB | 40 kB
+ "public"."fileset_ident" | 8192 bytes | 32 kB | 40 kB
+ "public"."webcapture_rev_release" | 8192 bytes | 32 kB | 40 kB
+ "public"."webcapture_ident" | 8192 bytes | 32 kB | 40 kB
+ "public"."fileset_rev" | 16 kB | 16 kB | 32 kB
+ "public"."webcapture_rev" | 16 kB | 16 kB | 32 kB
+ "public"."__diesel_schema_migrations" | 8192 bytes | 16 kB | 24 kB
+(41 rows)
generated by cgit v1.2.3 (git 2.39.1) at 2025-10-31 10:44:31 +0000