cgspace-notes/2023-12.md at 39d8d0876c25280f78bd16e86b3e2015ab1b0d70

alanorth/cgspace-notes

Fork 0

mirror of https://github.com/alanorth/cgspace-notes.git synced 2024-11-09 16:45:45 +01:00

Alan Orth 264cdcf1db

Add notes

2023-12-29 12:08:57 +03:00

11 KiB

Raw Blame History

title

date

author

2023-12-01

There is still high load on CGSpace and I don't know why
- I don't see a high number of sessions compared to previous days in the last few weeks

$ for file in dspace.log.2023-11-[23]*; do echo "$file"; grep -a -oE 'session_id=[A-Z0-9]{32}' "$file" | sort | uniq | wc -l; done
dspace.log.2023-11-20
22865
dspace.log.2023-11-21
20296
dspace.log.2023-11-22
19688
dspace.log.2023-11-23
17906
dspace.log.2023-11-24
18453
dspace.log.2023-11-25
17513
dspace.log.2023-11-26
19037
dspace.log.2023-11-27
21103
dspace.log.2023-11-28
23023
dspace.log.2023-11-29
23545
dspace.log.2023-11-30
21298

Even the number of unique IPs is not very high compared to the last week or so:

# awk '{print $1}' /var/log/nginx/{access,library-access,oai,rest}.log.1 | sort | uniq | wc -l
17023
# awk '{print $1}' /var/log/nginx/{access,library-access,oai,rest}.log.2.gz | sort | uniq | wc -l
17294
# awk '{print $1}' /var/log/nginx/{access,library-access,oai,rest}.log.3.gz | sort | uniq | wc -l
22057
# awk '{print $1}' /var/log/nginx/{access,library-access,oai,rest}.log.4.gz | sort | uniq | wc -l
32956
# awk '{print $1}' /var/log/nginx/{access,library-access,oai,rest}.log.5.gz | sort | uniq | wc -l
11415
# awk '{print $1}' /var/log/nginx/{access,library-access,oai,rest}.log.6.gz | sort | uniq | wc -l
15444
# awk '{print $1}' /var/log/nginx/{access,library-access,oai,rest}.log.7.gz | sort | uniq | wc -l
12648

It doesn't make any sense so I think I'm going to restart the server...
- After restarting the server the load went down to normal levels... who knows...
I started trying to see how I'm going to generate the fake statistics for the Alliance bitstream that was replaced
- I exported all the statistics for the owningItem now:

$ chrt -b 0 ./run.sh -s http://localhost:8081/solr/statistics -a export -o /tmp/stats-export.json -f 'owningItem:b5862bfa-9799-4167-b1cf-76f0f4ea1e18' -k uid

Importing them into DSpace Test didn't show the statistics in the Atmire module, but I see them in Solr...

2023-12-02

Export CGSpace to check for missing Initiative collection mappings
Start a harvest on AReS

2023-12-04

Send a message to Altmetric support because the item IWMI highlighted last month still doesn't show the attention score for the Handle after I tweeted it several times weeks ago
Spent some time writing a Python script to fix the literal MaxMind City JSON objects in our Solr statistics
- There are about 1.6 million of these, so I exported them using solr-import-export-json with the query city:com* but ended up finding many that have missing bundles, container bitstreams, etc:

city:com* AND -bundleName:[* TO *] AND -containerBitstream:[* TO *] AND -file_id:[* TO *] AND -owningItem:[* TO *] AND -version_id:[* TO *]

(Note the negation to find fields that are missing)
I don't know what I want to do with these yet

2023-12-05

I finished the fix_maxmind_stats.py script and fixed 1.6 million records and imported them on CGSpace after testing on DSpace 7 Test
Altmetric said there was a glitch regarding the Handle and DOI linking and they successfully re-scraped the item page and linked them
- They sent me a list of current production IPs and I notice that some of them are in our nginx bot network list:

$ for network in $(csvcut -c network /tmp/ips.csv | sed 1d | sort -u); do grepcidr $network ~/src/git/rmg-ansible-public/roles/dspace/files/nginx/bot-networks.conf; done
108.128.0.0/13 'bot';
46.137.0.0/16 'bot';
52.208.0.0/13 'bot';
52.48.0.0/13 'bot';
54.194.0.0/15 'bot';
54.216.0.0/14 'bot';
54.220.0.0/15 'bot';
54.228.0.0/15 'bot';
63.32.242.35/32     'bot';
63.32.0.0/14 'bot';
99.80.0.0/15 'bot'

I will remove those for now so that Altmetric doesn't have any unexpected issues harvesting

2023-12-08

Finalized the script to generate Solr statistics for Alliance research Mirjam
- The script is ilri/generate_solr_statistics.py
- I generated ~3,200 statistics based on her records of the download statistics of that item and imported them on CGSpace
Did some work on the DSpace 7 submission form
Peter asked for lists of affiliations, investors, and publishers to do some cleanups
- I generated a list from a CSV export instead of doing it based on a SQL dump...

$ csvcut -c 'cg.contributor.affiliation[en_US]' /tmp/initiatives.csv       \
  | sed -e 1d -e 's/^"//' -e 's/"$//' -e 's/||/\n/g' -e '/^$/d'            \
  | sort | uniq -c | sort -hr                                              \
  | awk 'BEGIN { FS = "^[[:space:]]+[[:digit:]]+[[:space:]]+" } {print $2}'\
  | sed -e '1i cg.contributor.affiliation' -e 's/^\(.*\)$/"\1"/'           \
  > /tmp/2023-12-08-initiatives-affiliations.csv

Export a list of authors as well:

localhost/dspace7= ☘ \COPY (SELECT DISTINCT text_value AS "dc.contributor.author", count(*) FROM metadatavalue WHERE dspace_object_id in (SELECT dspace_object_id FROM item) AND metadata_field_id = 3 GROUP BY "dc.contributor.author" ORDER BY count DESC) to /tmp/2023-12-08-authors.csv WITH CSV HEADER;
COPY 102435

2023-12-11

Work on OpenRXV dependencies and podman a bit
Peter noticed that the statistics for this month are very very low on CGSpace
- I don't know what is going on, perhaps it is related to me adjusting the nginx config last week?
- Ah, it's probably because of the spider patterns I updated on 2023-11

2023-12-16

Export CGSpace to check for missing Initiative collection mappings
Start a harvest on AReS

2023-12-17

Pull latest master branch for OpenRXV and deploy on the server
- I threw away some changes in the tree regarding the Angular base ref, and it broke AReS
- So note to self: we need to set the base ref in frontend/Dockerfile before building!
Now Salem fixed the country map

2023-12-18

Work a bit on the IFPRI-ISNAR archive from Leigh
More work on the DSpace 7 home page

2023-12-19

More work on the DSpace 7 home page
The Alliance TIP team is testing deposits to the DSpace 7 REST API and getting an HTTP 500 error
- In the DSpace logs I see this after they log in, create the item, and update the metadata:

2023-12-19 17:49:28,022 ERROR unknown unknown org.dspace.rest.Resource @ Something get wrong. Aborting context in finally statement.

I found some messages on the dspace-tech mailing list suggesting this might be an old bug: https://groups.google.com/g/dspace-tech/c/My1GUFYFGoU/m/tS7-WAJPAwAJ
- I restarted Tomcat and told the Alliance TIP team to try again

2023-12-20

The Alliance guys said that submitting via REST works now... sigh, so that's just some old DSpace 5/6 REST API bug
I lowercased all our AGROVOC keywords in dcterms.subject in SQL:

dspace=# BEGIN;
BEGIN
dspace=*# UPDATE metadatavalue SET text_value=LOWER(text_value) WHERE dspace_object_id IN (SELECT uuid FROM item) AND metadata_field_id=187 AND text_value ~ '[[:upper:]]';
UPDATE 462
dspace=*# COMMIT;
COMMIT

2023-12-25

Looking into Solr backups
- Since we are not running in Solr Cloud mode we need to use the replication endpoint for Solr standalone
- This works:

$ curl 'http://localhost:8983/solr/statistics/replication?command=backup'
{
  "responseHeader":{
    "status":0,
    "QTime":26},
  "status":"OK"}

Then I saw the size of the snapshot reach the size of the index...

# du -sh /var/solr/data/configsets/statistics/data/*
22G     /var/solr/data/configsets/statistics/data/index
16G     /var/solr/data/configsets/statistics/data/snapshot.20231225074111671
4.0K    /var/solr/data/configsets/statistics/data/snapshot_metadata
# du -sh /var/solr/data/configsets/statistics/data/*
22G     /var/solr/data/configsets/statistics/data/index
20G     /var/solr/data/configsets/statistics/data/snapshot.20231225074111671
4.0K    /var/solr/data/configsets/statistics/data/snapshot_metadata
# du -sh /var/solr/data/configsets/statistics/data/*
22G     /var/solr/data/configsets/statistics/data/index
21G     /var/solr/data/configsets/statistics/data/snapshot.20231225074111671
4.0K    /var/solr/data/configsets/statistics/data/snapshot_metadata
# du -sh /var/solr/data/configsets/statistics/data/*
22G     /var/solr/data/configsets/statistics/data/index
22G     /var/solr/data/configsets/statistics/data/snapshot.20231225074111671
4.0K    /var/solr/data/configsets/statistics/data/snapshot_metadata

Then I deleted the core and restored from the snapshot backup:

$ curl http://localhost:8983/solr/statistics/update -H "Content-type: text/xml" --data-binary '<delete><query>*:*</query></delete>'
$ curl http://localhost:8983/solr/statistics/update -H "Content-type: text/xml" --data-binary '<commit />'
$ curl 'http://localhost:8983/solr/statistics/replication?command=restore&name=statistics'

Interestingly the import worked fine, but created a new data index:

# du -sh /var/solr/data/configsets/statistics/data/*
4.0K    /var/solr/data/configsets/statistics/data/index.properties
22G     /var/solr/data/configsets/statistics/data/restore.20231225154626463
4.0K    /var/solr/data/configsets/statistics/data/snapshot_metadata
22G     /var/solr/data/configsets/statistics/data/snapshot.statistics

Not sure the implications of that—Solr uses the data just fine
I can surely use this for atomic Solr backups

2023-12-27

Delete duplicate metadata as described in my DSpace issue from last year: https://github.com/DSpace/DSpace/issues/8253
Do some other metadata cleanups on CGSpace
- I also looked up our DOIs on Crossref to get some missing abstracts and correct licenses and dates
Some minor work on the CGSpace DSpace 7 theme to fix the navbar on mobile
Some work on the IFPRI ISNAR archive

2023-12-28

I started porting the cgspace-java-helpers to DSpace 7
Some work on the IFPRI ISNAR archive
- I ended up going through most of the PDFs to get better dates and abstracts

2023-12-29

I created a new Hetzner server to replace the current DSpace 6 CGSpace next week when we migrate to DSpace 7
Interesting, I haven't checked for content pointing to legacy domains in several years (!)
- inurl:mahider.cgiar.org: 0 results on Google!
- inurl:mahider.ilri.org: 2,100 results on Google
- inurl:mahider.ilri.org inurl:https: 2 results on Google (!)
- inurl:dspace.ilri.org: 1,390 results on Google
- inurl:dspace.ilri.org inurl:https: 0 results on Google (!)
So it seems I can do away with the HTTPS virtual hosts finally
- Well my current certificates expired on 2021-02-13 and nobody noticed... so...

11 KiB Raw Blame History

2023-12-01

2023-12-02

2023-12-04

2023-12-05

2023-12-08

2023-12-11

2023-12-16

2023-12-17

2023-12-18

2023-12-19

2023-12-20

2023-12-25

2023-12-27

2023-12-28

2023-12-29

11 KiB

Raw Blame History