cgspace-notes/2020-02.md at 8d71539431e1b8c983e8611e8c56d89e735e89f7

alanorth/cgspace-notes

Fork 0

mirror of https://github.com/alanorth/cgspace-notes.git synced 2024-11-17 20:27:05 +01:00

Alan Orth 8d71539431

Update notes for 2020-02-09

2020-02-09 15:55:49 +02:00

23 KiB

Raw Blame History

title

date

author

2020-02-02

Continue working on porting CGSpace's DSpace 5 code to DSpace 6.3 that I started yesterday
- Sign up for an account with MaxMind so I can get the GeoLite2-City.mmdb database
- I still need to wire up the API credentials and cron job into the Ansible infrastructure playbooks
- Fix some minor issues in the config and XMLUI themes, like removing Atmire stuff
- The code finally builds and runs with a fresh install

Now we don't specify the build environment because site modification are in local.cfg, so we just build like this:

$ schedtool -D -e ionice -c2 -n7 nice -n19 mvn -U -Dmirage2.on=true -Dmirage2.deps.included=false clean package

And it seems that we need to enabled pg_crypto now (used for UUIDs):

$ psql -h localhost -U postgres dspace63
dspace63=# CREATE EXTENSION pgcrypto;
CREATE EXTENSION pgcrypto;

I tried importing a PostgreSQL snapshot from CGSpace and had errors due to missing Atmire database migrations
- If I try to run dspace database migrate I get the IDs of the migrations that are missing
- I delete them manually in psql:

dspace63=# DELETE FROM schema_version WHERE version IN ('5.0.2015.01.27', '5.6.2015.12.03.2', '5.6.2016.08.08', '5.0.2017.04.28', '5.0.2017.09.25', '5.8.2015.12.03.3');

Then I ran dspace database migrate and got an error:

$ ~/dspace63/bin/dspace database migrate

Database URL: jdbc:postgresql://localhost:5432/dspace63?ApplicationName=dspaceCli
Migrating database to latest version... (Check dspace logs for details)
Migration exception:
java.sql.SQLException: Flyway migration error occurred
        at org.dspace.storage.rdbms.DatabaseUtils.updateDatabase(DatabaseUtils.java:673)
        at org.dspace.storage.rdbms.DatabaseUtils.updateDatabase(DatabaseUtils.java:576)
        at org.dspace.storage.rdbms.DatabaseUtils.main(DatabaseUtils.java:221)
        at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
        at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
        at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
        at java.lang.reflect.Method.invoke(Method.java:498)
        at org.dspace.app.launcher.ScriptLauncher.runOneCommand(ScriptLauncher.java:229)
        at org.dspace.app.launcher.ScriptLauncher.main(ScriptLauncher.java:81)
Caused by: org.flywaydb.core.internal.dbsupport.FlywaySqlScriptException:
Migration V6.0_2015.03.07__DS-2701_Hibernate_migration.sql failed
-----------------------------------------------------------------
SQL State  : 2BP01
Error Code : 0
Message    : ERROR: cannot drop table metadatavalue column resource_id because other objects depend on it
  Detail: view eperson_metadata depends on table metadatavalue column resource_id
  Hint: Use DROP ... CASCADE to drop the dependent objects too.
Location   : org/dspace/storage/rdbms/sqlmigration/postgres/V6.0_2015.03.07__DS-2701_Hibernate_migration.sql (/home/aorth/src/git/DSpace-6.3/file:/home/aorth/dspace63/lib/dspace-api-6.3.jar!/org/dspace/storage/rdbms/sqlmigration/postgres/V6.0_2015.03.07__DS-2701_Hibernate_migration.sql)
Line       : 391
Statement  : ALTER TABLE metadatavalue DROP COLUMN IF EXISTS resource_id

        at org.flywaydb.core.internal.dbsupport.SqlScript.execute(SqlScript.java:117)
        at org.flywaydb.core.internal.resolver.sql.SqlMigrationExecutor.execute(SqlMigrationExecutor.java:71)
        at org.flywaydb.core.internal.command.DbMigrate.doMigrate(DbMigrate.java:352)
        at org.flywaydb.core.internal.command.DbMigrate.access$1100(DbMigrate.java:47)
        at org.flywaydb.core.internal.command.DbMigrate$4.doInTransaction(DbMigrate.java:308)
        at org.flywaydb.core.internal.util.jdbc.TransactionTemplate.execute(TransactionTemplate.java:72)
        at org.flywaydb.core.internal.command.DbMigrate.applyMigration(DbMigrate.java:305)
        at org.flywaydb.core.internal.command.DbMigrate.access$1000(DbMigrate.java:47)
        at org.flywaydb.core.internal.command.DbMigrate$2.doInTransaction(DbMigrate.java:230)
        at org.flywaydb.core.internal.command.DbMigrate$2.doInTransaction(DbMigrate.java:173)
        at org.flywaydb.core.internal.util.jdbc.TransactionTemplate.execute(TransactionTemplate.java:72)
        at org.flywaydb.core.internal.command.DbMigrate.migrate(DbMigrate.java:173)
        at org.flywaydb.core.Flyway$1.execute(Flyway.java:959)
        at org.flywaydb.core.Flyway$1.execute(Flyway.java:917)
        at org.flywaydb.core.Flyway.execute(Flyway.java:1373)
        at org.flywaydb.core.Flyway.migrate(Flyway.java:917)
        at org.dspace.storage.rdbms.DatabaseUtils.updateDatabase(DatabaseUtils.java:662)
        ... 8 more
Caused by: org.postgresql.util.PSQLException: ERROR: cannot drop table metadatavalue column resource_id because other objects depend on it
  Detail: view eperson_metadata depends on table metadatavalue column resource_id
  Hint: Use DROP ... CASCADE to drop the dependent objects too.
        at org.postgresql.core.v3.QueryExecutorImpl.receiveErrorResponse(QueryExecutorImpl.java:2422)
        at org.postgresql.core.v3.QueryExecutorImpl.processResults(QueryExecutorImpl.java:2167)
        at org.postgresql.core.v3.QueryExecutorImpl.execute(QueryExecutorImpl.java:306)
        at org.postgresql.jdbc.PgStatement.executeInternal(PgStatement.java:441)
        at org.postgresql.jdbc.PgStatement.execute(PgStatement.java:365)
        at org.postgresql.jdbc.PgStatement.executeWithFlags(PgStatement.java:307)
        at org.postgresql.jdbc.PgStatement.executeCachedSql(PgStatement.java:293)
        at org.postgresql.jdbc.PgStatement.executeWithFlags(PgStatement.java:270)
        at org.postgresql.jdbc.PgStatement.execute(PgStatement.java:266)
        at org.apache.commons.dbcp2.DelegatingStatement.execute(DelegatingStatement.java:291)
        at org.apache.commons.dbcp2.DelegatingStatement.execute(DelegatingStatement.java:291)
        at org.flywaydb.core.internal.dbsupport.JdbcTemplate.executeStatement(JdbcTemplate.java:238)
        at org.flywaydb.core.internal.dbsupport.SqlScript.execute(SqlScript.java:114)
        ... 24 more

I think I might need to update the sequences first... nope
Perhaps it's due to some missing bitstream IDs and I need to run dspace cleanup on CGSpace and take a new PostgreSQL dump... nope
A thread on the dspace-tech mailing list regarding this migration noticed that his database had some views created that were using the resource_id column
Our database had the same issue, where the eperson_metadata view was created by something (Atmire module?) but has no references in the vanilla DSpace code, so I dropped it and tried the migration again:

dspace63=# DROP VIEW eperson_metadata;
DROP VIEW

After that the migration was successful and DSpace starts up successfully and begins indexing
- xmlui, solr, jspui, rest, and oai are working (rest was redirecting to HTTPS, so I set the Tomcat connector to secure="true" and it fixed it on localhost, but caused other issues so I disabled it for now)
- I started diffing our themes against the Mirage 2 reference theme to capture the latest changes

2020-02-03

Update DSpace mimetype fallback images from KDE Breeze Icons project
- Our icons are four years old (see my bitstream icons demo)
Issues remaining in the DSpace 6 port of our CGSpace 5.x code:
- Community and collection pages only show one recent submission (seems that there is only one item in Solr?)
- Community and collection pages have tons of "Browse" buttons that we need to remove
- Order of navigation elements in right side bar ("My Account" etc, compare to DSpace Test)
- Home page trail says "CGSpace Home" instead of "CGSpace Home / Community List" (see DSpace Test)
There are lots of errors in the DSpace log, which might explain some of the issues with recent submissions / Solr:

2020-02-03 10:27:14,485 ERROR org.dspace.browse.ItemCountDAOSolr @ caught exception: 
org.dspace.discovery.SearchServiceException: Invalid UUID string: 1
2020-02-03 13:20:20,475 ERROR org.dspace.app.xmlui.aspect.discovery.AbstractRecentSubmissionTransformer @ Caught SearchServiceException while retrieving recent submission for: home page
org.dspace.discovery.SearchServiceException: Invalid UUID string: 111210

If I look in Solr's search core I do actually see items with integers for their resource ID, which I think are all supposed to be UUIDs now...
I dropped all the documents in the search core:

$ http --print b 'http://localhost:8080/solr/search/update?stream.body=<delete><query>*:*</query></delete>&commit=true'

Still didn't work, so I'm going to try a clean database import and migration:

$ createdb -h localhost -U postgres -O dspacetest --encoding=UNICODE dspace63
$ psql -h localhost -U postgres -c 'alter user dspacetest superuser;'
$ pg_restore -h localhost -U postgres -d dspace63 -O --role=dspacetest -h localhost dspace_2020-01-27.backup
$ psql -h localhost -U postgres -c 'alter user dspacetest nosuperuser;'
$ psql -h localhost -U postgres dspace63                               
dspace63=# CREATE EXTENSION pgcrypto;
dspace63=# DELETE FROM schema_version WHERE version IN ('5.0.2015.01.27', '5.6.2015.12.03.2', '5.6.2016.08.08', '5.0.2017.04.28', '5.0.2017.09.25', '5.8.2015.12.03.3');
dspace63=# DROP VIEW eperson_metadata;
dspace63=# \q
$ psql -h localhost -U postgres -f ~/src/git/DSpace/dspace/etc/postgres/update-sequences.sql dspace63
$ ~/dspace63/bin/dspace database migrate

I notice that the indexing doesn't work correctly if I start it manually with dspace index-discovery -b (search.resourceid becomes an integer!)
- If I induce an indexing by touching dspace/solr/search/conf/reindex.flag the search.resourceid are all UUIDs...
Speaking of database stuff, there was a performance-related update for the indexes that we used in DSpace 5
- We might want to apply it in DSpace 6, as it was never merged to 6.x, but it helped with the performance of /submissions in XMLUI for us in [2018-03]({{< relref path="2018-03.md" >}})

2020-02-04

The indexing issue I was having yesterday seems to only present itself the first time a new installation is running DSpace 6
- Once the indexing induced by touching dspace/solr/search/conf/reindex.flag has finished, subsequent manual invocations of dspace index-discovery -b work as expected
- Nevertheless, I sent a message to the dspace-tech mailing list describing the issue to see if anyone has any comments
I am seeing that the number of important commits on the unreleased DSpace 6.4 are really numerous and it might be better for us to target that version
- I did a simple test and it's easy to rebase my current 6.3 branch on top of the upstream dspace-6_x branch:

$ git checkout -b 6_x-dev64 6_x-dev
$ git rebase -i upstream/dspace-6_x

I finally understand why our themes show all the "Browse by" buttons on community and collection pages in DSpace 6.x
- The code in ./dspace-xmlui/src/main/java/org/dspace/app/xmlui/aspect/browseArtifacts/CommunityBrowse.java iterates over all the browse indexes and prints them when it is called
- The XMLUI theme code in dspace/modules/xmlui-mirage2/src/main/webapp/themes/0_CGIAR/xsl/preprocess/browse.xsl calls the template because the id of the div matches "aspect.browseArtifacts.CommunityBrowse.list.community-browse"
- I checked the DRI of a community page on my local 6.x and DSpace Test 5.x by appending ?XML to the URL and I see the ID is missing on DSpace 5.x
- The issue is the same with the ordering of the "My Account" link, but in Navigation.java
- I tried modifying preprocess/browse.xsl but it always ends up printing some default list of browse by links...
- I'm starting to wonder if Atmire's modules somehow override this, as I don't see how CommunityBrowse.java can behave like ours on DSpace 5.x unless they have overridden it (as the open source code is the same in 5.x and 6.x)
- At least the "account" link in the sidebar is overridden in our 5.x branch because Atmire copied a modified Navigation.java to the local xmlui modules folder... so that explains that (and it's easy to replicate in 6.x)

2020-02-05

UptimeRobot told me that AReS Explorer crashed last night, so I logged into it, ran all updates, and rebooted it
Testing Discovery indexing speed on my local DSpace 6.3:

$ time schedtool -D -e ~/dspace63/bin/dspace index-discovery -b
schedtool -D -e ~/dspace63/bin/dspace index-discovery -b  3771.78s user 93.63s system 41% cpu 2:34:19.53 total
schedtool -D -e ~/dspace63/bin/dspace index-discovery -b  3360.28s user 82.63s system 38% cpu 2:30:22.07 total
schedtool -D -e ~/dspace63/bin/dspace index-discovery -b  4678.72s user 138.87s system 42% cpu 3:08:35.72 total
schedtool -D -e ~/dspace63/bin/dspace index-discovery -b  3334.19s user 86.54s system 35% cpu 2:41:56.73 total

DSpace 5.8 was taking about 1 hour (or less on this laptop), so this is 2-3 times longer!

$ time schedtool -D -e ~/dspace/bin/dspace index-discovery -b
schedtool -D -e ~/dspace/bin/dspace index-discovery -b  299.53s user 69.67s system 20% cpu 30:34.47 total
schedtool -D -e ~/dspace/bin/dspace index-discovery -b  270.31s user 69.88s system 19% cpu 29:01.38 total

Checking out the DSpace 6.x REST API query client
- There is a tutorial that explains how it works and I see it is very powerful because you can export a CSV of results in order to fix and re-upload them with batch import!
- Custom queries can be added in dspace-rest/src/main/webapp/static/reports/restQueryReport.js
I noticed two new bots in the logs with the following user agents:
- Jersey/2.6 (HttpUrlConnection 1.8.0_152)
- magpie-crawler/1.1 (U; Linux amd64; en-GB; +http://www.brandwatch.net)
I filed an issue to add Jersey to the COUNTER-Robots list
Peter noticed that the statlets on community, collection, and item pages aren't working on CGSpace
- I thought it might be related to the fact that the yearly sharding didn't complete successfully this year so the statistics-2019 core is empty
- I removed the statistics-2019 core and had to restart Tomcat like six times before all cores would load properly (ugh!!!!)
- After that the statlets were working properly...
Run all system updates on DSpace Test (linode19) and restart it

2020-02-06

I sent a mail to the dspace-tech mailing list asking about slow Discovery indexing speed in DSpace 6
I destroyed my PostgreSQL 9.6 containers and re-created them using PostgreSQL 10 to see if there are any speedups with DSpace 6.x:

$ podman pull postgres:10-alpine
$ podman run --name dspacedb10 -v dspacedb_data:/var/lib/postgresql/data -e POSTGRES_PASSWORD=postgres -p 5432:5432 -d postgres:10-alpine
$ createuser -h localhost -U postgres --pwprompt dspacetest
$ createdb -h localhost -U postgres -O dspacetest --encoding=UNICODE dspacetest
$ createdb -h localhost -U postgres -O dspacetest --encoding=UNICODE dspace63
$ psql -h localhost -U postgres -c 'alter user dspacetest superuser;'
$ pg_restore -h localhost -U postgres -d dspacetest -O --role=dspacetest -h localhost ~/Downloads/cgspace_2020-02-06.backup
$ pg_restore -h localhost -U postgres -d dspace63 -O --role=dspacetest -h localhost ~/Downloads/cgspace_2020-02-06.backup
$ psql -h localhost -U postgres -f ~/src/git/DSpace/dspace/etc/postgres/update-sequences.sql dspacetest
$ psql -h localhost -U postgres -f ~/src/git/DSpace/dspace/etc/postgres/update-sequences.sql dspace63
$ psql -h localhost -U postgres -c 'alter user dspacetest nosuperuser;'
$ psql -h localhost -U postgres dspace63                               
dspace63=# CREATE EXTENSION pgcrypto;
dspace63=# DELETE FROM schema_version WHERE version IN ('5.0.2015.01.27', '5.6.2015.12.03.2', '5.6.2016.08.08', '5.0.2017.04.28', '5.0.2017.09.25', '5.8.2015.12.03.3');
dspace63=# DROP VIEW eperson_metadata;
dspace63=# \q

I purged ~33,000 hits from the "Jersey/2.6" bot in CGSpace's statistics using my check-spider-hits.sh script:

$ ./check-spider-hits.sh -d -p -f /tmp/jersey -s statistics -u http://localhost:8081/solr
$ for year in 2018 2017 2016 2015; do ./check-spider-hits.sh -d -p -f /tmp/jersey -s "statistics-${year}" -u http://localhost:8081/solr; done

I noticed another user agen in the logs that we should add to the list:

ReactorNetty/0.9.2.RELEASE

I made an issue on the COUNTER-Robots repository
I found a nice tool for exporting and importing Solr records and it seems to workfor exporting our 2019 stats from the large statistics core!

$ ./run.sh -s http://localhost:8081/solr/statistics -a export -o /tmp/statistics-2019-01.json -f 'dateYearMonth:2019-01' -k uid
$ ls -lh /tmp/statistics-2019-01.json
-rw-rw-r-- 1 aorth aorth 3.7G Feb  6 09:26 /tmp/statistics-2019-01.json

Then I tested importing this by creating a new core in my development environment:

$ curl 'http://localhost:8080/solr/admin/cores?action=CREATE&name=statistics-2019&instanceDir=/home/aorth/dspace/solr/statistics&dataDir=/home/aorth/dspace/solr/statistics-2019/data'
$ ./run.sh -s http://localhost:8080/solr/statistics-2019 -a import -o ~/Downloads/statistics-2019-01.json -k uid

This imports the records into the core, but DSpace can't see them, and when I restart Tomcat the core is not seen by Solr...
I got the core to load by adding it to dspace/solr/solr.xml manually, ie:

  <cores adminPath="/admin/cores">
  ...
    <core name="statistics" instanceDir="statistics" />
    <core name="statistics-2019" instanceDir="statistics">
        <property name="dataDir" value="/home/aorth/dspace/solr/statistics-2019/data" />
    </core>
  ...
  </cores>

But I don't like having to do that... why doesn't it load automatically?
I sent a mail to the dspace-tech mailing list to ask about it
Just for fun I tried to load these stats into a Solr 7.7.2 instance using the DSpace 7 solr config:
First, create a Solr statistics core using the DSpace 7 config:

$ ./bin/solr create_core -c statistics -d ~/src/git/DSpace/dspace/solr/statistics/conf -p 8983

Then try to import the stats, skipping a shitload of fields that are apparently added to our Solr statistics by Atmire modules:

$ ./run.sh -s http://localhost:8983/solr/statistics -a import -o ~/Downloads/statistics-2019-01.json -k uid -S author_mtdt,author_mtdt_search,iso_mtdt_search,iso_mtdt,subject_mtdt,subject_mtdt_search,containerCollection,containerCommunity,containerItem,countryCode_ngram,countryCode_search,cua_version,dateYear,dateYearMonth,geoipcountrycode,ip_ngram,ip_search,isArchived,isInternal,isWithdrawn,containerBitstream,file_id,referrer_ngram,referrer_search,userAgent_ngram,userAgent_search,version_id,complete_query,complete_query_search,filterquery,ngram_query_search,ngram_simplequery_search,simple_query,simple_query_search,range,rangeDescription,rangeDescription_ngram,rangeDescription_search,range_ngram,range_search,actingGroupId,actorMemberGroupId,bitstreamCount,solr_update_time_stamp,bitstreamId

OK that imported! I wonder if it works... maybe I'll try another day

2020-02-07

I did some investigation into DSpace indexing performance using flame graphs
- Excellent introduction: http://www.brendangregg.com/flamegraphs.html
- Using flame graphs with java: https://netflixtechblog.com/java-in-flames-e763b3d32166
- Fantastic wrapper scripts for doing perf on Java processes: https://github.com/jvm-profiling-tools/perf-map-agent

$ cd ~/src/git/perf-map-agent
$ cmake     .
$ make
$ ./bin/create-links-in ~/.local/bin
$ export FLAMEGRAPH_DIR=/home/aorth/src/git/FlameGraph
$ export JAVA_HOME=/usr/lib/jvm/java-8-openjdk
$ export JAVA_OPTS="-XX:+PreserveFramePointer"
$ ~/dspace63/bin/dspace index-discovery -b &
# pid of tomcat java process
$ perf-java-flames 4478
# pid of java indexing process
$ perf-java-flames 11359

All Java processes need to have -XX:+PreserveFramePointer if you want to trace their methods
I did the same tests against DSpace 5.8 and 6.4-SNAPSHOT's CLI indexing process and Tomcat process
- For what it's worth, it appears all the Hibernate stuff is in the CLI processes, so we don't need to trace the Tomcat process
Here is the flame graph for DSpace 5.8's dspace index-discovery -b java process:

Here is the flame graph for DSpace 6.4-SNAPSHOT's dspace index-discovery -b java process:

If the width of the stacks indicates time, then it's clear that Hibernate takes longer...
Apparently there is a "flame diff" tool, I wonder if we can use that to compare!

2020-02-09

This weekend I did a lot more testing of indexing performance with our DSpace 5.8 branch, vanilla DSpace 5.10, and vanilla DSpace 6.4-SNAPSHOT:

# CGSpace 5.8
schedtool -D -e ~/dspace/bin/dspace index-discovery -b  385.72s user 131.16s system 19% cpu 43:21.18 total
schedtool -D -e ~/dspace/bin/dspace index-discovery -b  382.95s user 127.31s system 20% cpu 42:10.07 total
schedtool -D -e ~/dspace/bin/dspace index-discovery -b  368.56s user 143.97s system 20% cpu 42:22.66 total
schedtool -D -e ~/dspace/bin/dspace index-discovery -b  360.09s user 104.03s system 19% cpu 39:24.41 total

# Vanilla DSpace 5.10
schedtool -D -e ~/dspace510/bin/dspace index-discovery -b  236.19s user 59.70s system 3% cpu 2:03:31.14 total
schedtool -D -e ~/dspace510/bin/dspace index-discovery -b  232.41s user 50.38s system 3% cpu 2:04:16.00 total

# Vanilla DSpace 6.4-SNAPSHOT
schedtool -D -e ~/dspace63/bin/dspace index-discovery -b  5112.96s user 127.80s system 40% cpu 3:36:53.98 total
schedtool -D -e ~/dspace63/bin/dspace index-discovery -b  5112.96s user 127.80s system 40% cpu 3:21:0.0 total

I generated better flame graphs for the DSpace indexing process by using perf-record-stack and filtering out the java process:

$ export JAVA_HOME=/usr/lib/jvm/java-8-openjdk
$ export PERF_RECORD_SECONDS=60
$ export JAVA_OPTS="-XX:+PreserveFramePointer"
$ time schedtool -D -e ~/dspace/bin/dspace index-discovery -b &
# process id of java indexing process (not Tomcat)
$ perf-java-record-stack 169639
$ sudo perf script -i /tmp/perf-169639.data > out.dspace510-1
$ cat out.dspace510-1 | ../FlameGraph/stackcollapse-perf.pl | grep -E '^java' | ../FlameGraph/flamegraph.pl --color=java --hash > out.dspace510-1.svg

All data recorded on my laptop with the same kernel, same boot, etc.
CGSpace 5.8 (with Atmire patches):

Vanilla DSpace 5.10:

Vanilla DSpace 6.4-SNAPSHOT:

I sent my feedback to the dspace-tech mailing list so someone can hopefully comment.

23 KiB Raw Blame History

2020-02-02

2020-02-03

2020-02-04

2020-02-05

2020-02-06

2020-02-07

2020-02-09

23 KiB

Raw Blame History