ARXIVNG-1519 tuning search API for production

erickpeirson · erickpeirson · commit 4129fbbc1e78 · 2019-02-25T16:13:07.000-05:00
diff --git a/Dockerfile-api b/Dockerfile-api
@@ -3,7 +3,7 @@
 # Defines the runtime for the arXiv search API, which provides a metadata
 # query API backed by Elasticsearch.
 
-FROM arxiv/base:0.12.1
+FROM arxiv/base:0.14.3
 
 WORKDIR /opt/arxiv
 
@@ -22,11 +22,7 @@ ENV PATH "/opt/arxiv:${PATH}"
 ADD schema /opt/arxiv/schema
 ADD mappings /opt/arxiv/mappings
 ADD search /opt/arxiv/search
-ADD wsgi-api.py /opt/arxiv/wsgi.py
-RUN pip install uwsgi
-
-ADD bin/start_search.sh /opt/arxiv/
-RUN chmod +x /opt/arxiv/start_search.sh
+ADD wsgi-api.py config/uwsgi-api.ini /opt/arxiv/
 
 ENV LC_ALL en_US.utf8
 ENV LANG en_US.utf8
@@ -44,16 +40,5 @@ ENV METADATA_ENDPOINT https://arxiv.org/docmeta_bulk/
 
 EXPOSE 8000
 
-#CMD /bin/bash
-ENTRYPOINT ["/opt/arxiv/start_search.sh"]
-CMD ["--http-socket", ":8000", \
-     "-M", \
-     "-t 3000", \
-     "--manage-script-name", \
-     "--processes", "8", \
-     "--threads", "1", \
-     "--async", "100", \
-     "--ugreen", \
-     "--buffer-size", "65535", \
-     "--mount", "/metadata=wsgi.py", \
-     "--logformat", "%(addr) %(addr) - %(user_id)|%(session_id) [%(rtime)] [%(uagent)] \"%(method) %(uri) %(proto)\" %(status) %(size) %(micros) %(ttfb)"]
+ENTRYPOINT ["pipenv", "run"]
+CMD ["uwsgi", "--ini", "/opt/arxiv/uwsgi-api.ini"]
diff --git a/Pipfile b/Pipfile
@@ -4,7 +4,7 @@ verify_ssl = true
 name = "pypi"
 
 [packages]
-arxiv-auth = "==0.2.3"
+arxiv-auth = "==0.2.7"
 arxiv-base = "==0.12.1"
 boto = "==2.48.0"
 "boto3" = "==1.6.6"
diff --git a/Pipfile.lock b/Pipfile.lock
diff --git a/config/uwsgi-api.ini b/config/uwsgi-api.ini
@@ -0,0 +1,14 @@
+[uwsgi]
+http-socket = :8000
+chdir = /opt/arxiv/
+wsgi-file = wsgi-api.py
+callable = application
+master = true
+harakiri = 3000
+manage-script-name = true
+processes = 1
+queue = 0
+threads = 1
+single-interpreter = true
+mount = /=wsgi-api.py
+logformat = "%(addr) %(addr) - %(user_id)|%(session_id) [%(rtime)] [%(uagent)] \"%(method) %(uri) %(proto)\" %(status) %(size) %(micros) %(ttfb)"
diff --git a/search/routes/api/serialize.py b/search/routes/api/serialize.py
@@ -1,7 +1,7 @@
 """Serializers for API responses."""
 
 from typing import Union, Optional
-from lxml import etree
+from xml.etree import ElementTree as etree
 from flask import jsonify, url_for
 
 from arxiv import status
