The signature of ImageService.storeImageBytes is

Image storeImageBytes(byte[] bytes, String originalFilename, long filesize, String contentType,
String uploaderId, Map metadata)

However, a number of calls to this method don't contain the final metadata Map parameter and so crash at runtime without informing the user that the upload failed or writing the error the log.

I will implement a PR when I get the chance, because without them the eFlora reports won't look pretty :-)
/cc @djtfmartin @m-r-c

The workflow for navigating to linked occurrence records is not intuitive and some users are finding it easier to use the support button to report misidentified species from images.ala.org.au, which is not ideal.

Refs tickets 25350 and 25351 in helpdesk

Some thing weird about the attached image, ecodata accepts it happily, but image server doesn't like it!)

Example: https://images.ala.org.au/image/details/202813734

Note from Chris:
[11:34 AM] Chris Godwin: It's might the colour profile
[11:34 AM] Chris Godwin: I had to add an extension library to ImageIO on ecodata to handle a few extra image types

The description in meta tag should be configurable with the name of others nodes:

Or see source of https://imagenes.gbif.es

Most of the URLs in this app are under the authenticateOnlyIfLoggedInFilterPattern filter. This can result in some users seeing missing image icons in apps that show images out of the image-service.

I am able to replicate this issue in the biocache-hubs by doing the following:

• Visit a page with images, e.g.http://biocache.ala.org.au/occurrences/search?q=pink#tab_recordImages
• logout (if logged in)
• login again but DO NOT CLICK the "Remember me" tick box
• Wait 24 hours (for session to expire) but leave page up in browser
• refresh page - image will be missing (see screen shot)

Suggested fix is to ensure all public URLs are not under CAS at all. Note that URLs under the authenticateOnlyIfLoggedInFilterPattern will not be excluded is included in the uriExclusionFilter (it only works against the uriFilterPattern - this is a bug/feature).

In Spain, we have a lot of broken images, mainly because people share html/php instead of images. For instance:
http://161.111.171.57/herbarioV/visorVCat.php?img=MA-01-00020718
is a php visor of images, so, our image service obviously cannot process this.

Here an occurrence with a broken image and a correct image:
https://registros.gbif.es/occurrences/ce778852-8e22-4866-8f96-c4ef427cf0a2

Other broken images, like:
https://images.ala.org.au/image/details/252342823
with size 0x0.

So what I did, if this serve as inspiration now that the image service is under improvement, is to add a default 404 broken image with some apache directive like:

ErrorDocument 404 /store/broken.png

I also used ProxyErrorOverride On to show the same broken image on 404 errors in tomcat.

More info:
https://stackoverflow.com/questions/4017470/replace-invalid-image-url-with-404-image
The broken image we use has CC license:
https://thenounproject.com/term/broken-image/389560/

Just to comment, please feel free to close this if does not fit with the image service development.

request for bulk koala images to train an image recognition algorithm couldn't be done at the moment.

any bulk image download has to include licensing and attribution (should be in the XIF)

Load testing on image-service-1.0.5 shows that it consistently runs out of file descriptors at a particular query rate in our current images-beta setup, about 140 requests per second, which is not unreasonable in the long term that it would be hit outside of load testing given we show 20 images at a time to biocache users, and we want to load images at high capacity.

This could be improved by using a pooled or limited HTTP client to request the /store/ URLs, where the JVM goes out and back in through nginx to fetch the images, rather than accessing them directly.

The relevant code where it looks like this can be improved for a large number of cases at once is the following:

Images fails to connect to postgres after a package upgrade and shows the following error in its log files:

2018-06-05 06:43:45,679 [quartzScheduler_Worker-2] ERROR util.JDBCExceptionReporter  - FATAL: terminating connection due to administrator command
2018-06-05 06:43:45,679 [quartzScheduler_Worker-2] ERROR util.JDBCExceptionReporter  - An I/O error occurred while sending to the backend.
2018-06-05 06:43:45,680 [quartzScheduler_Worker-2] ERROR util.JDBCExceptionReporter  - This connection has been closed.
2018-06-05 06:43:45,681 [quartzScheduler_Worker-2] INFO  images.LogService  - Exception thrown in job handler
2018-06-05 06:43:45,688 [quartzScheduler_Worker-2] INFO  images.LogService  - Error: could not inspect JDBC autocommit mode; nested exception is org.hibernate.exception.JDBCConnectionException: could not inspect JDBC autocommit mode
2018-06-05 06:43:45,688 [quartzScheduler_Worker-2] INFO  core.JobRunShell  - Job GRAILS_JOBS.au.org.ala.images.ProcessBackgroundTasksJob threw a JobExecutionException: 
org.quartz.JobExecutionException: org.springframework.dao.DataAccessResourceFailureException: could not inspect JDBC autocommit mode; nested exception is org.hibernate.exception.JDBCConnectionException: could not inspect JDBC autocommit mode [See nested exception: org.springframework.dao.DataAccessResourceFailureException: could not inspect JDBC autocommit mode; nested exception is org.hibernate.exception.JDBCConnectionException: could not inspect JDBC autocommit mode]
        at grails.plugins.quartz.GrailsJobFactory$GrailsJob.execute(GrailsJobFactory.java:111)
        at org.quartz.core.JobRunShell.run(JobRunShell.java:202)
        at org.quartz.simpl.SimpleThreadPool$WorkerThread.run(SimpleThreadPool.java:573)
Caused by: org.springframework.dao.DataAccessResourceFailureException: could not inspect JDBC autocommit mode; nested exception is org.hibernate.exception.JDBCConnectionException: could not inspect JDBC autocommit mode
        at org.grails.datastore.gorm.GormStaticApi$_methodMissing_closure2.doCall(GormStaticApi.groovy:101)
        at au.org.ala.images.SettingService.getOrCreateSetting(SettingService.groovy:109)
        at au.org.ala.images.SettingService.getSettingFromStack(SettingService.groovy:85)
        at au.org.ala.images.SettingService.getBoolSetting(SettingService.groovy:93)
        at au.org.ala.images.SettingService.getBackgroundTasksEnabled(SettingService.groovy:25)
        at au.org.ala.images.ProcessBackgroundTasksJob.execute(ProcessBackgroundTasksJob.groovy:17)
        at grails.plugins.quartz.GrailsJobFactory$GrailsJob.execute(GrailsJobFactory.java:104)
        ... 2 more
Caused by: org.hibernate.exception.JDBCConnectionException: could not inspect JDBC autocommit mode
        ... 9 more
Caused by: org.postgresql.util.PSQLException: This connection has been closed.
        at org.postgresql.jdbc2.AbstractJdbc2Connection.checkClosed(AbstractJdbc2Connection.java:820)
        at org.postgresql.jdbc2.AbstractJdbc2Connection.getAutoCommit(AbstractJdbc2Connection.java:781)
        ... 9 more
2018-06-05 06:43:45,689 [quartzScheduler_Worker-2] ERROR listeners.ExceptionPrinterJobListener  - Exception occurred in job: Grails Job
org.quartz.JobExecutionException: org.springframework.dao.DataAccessResourceFailureException: could not inspect JDBC autocommit mode; nested exception is org.hibernate.exception.JDBCConnectionException: could not inspect JDBC autocommit mode [See nested exception: org.springframework.dao.DataAccessResourceFailureException: could not inspect JDBC autocommit mode; nested exception is org.hibernate.exception.JDBCConnectionException: could not inspect JDBC autocommit mode]
        at grails.plugins.quartz.GrailsJobFactory$GrailsJob.execute(GrailsJobFactory.java:111)
        at org.quartz.core.JobRunShell.run(JobRunShell.java:202)
        at org.quartz.simpl.SimpleThreadPool$WorkerThread.run(SimpleThreadPool.java:573)
Caused by: org.springframework.dao.DataAccessResourceFailureException: could not inspect JDBC autocommit mode; nested exception is org.hibernate.exception.JDBCConnectionException: could not inspect JDBC autocommit mode
        at org.grails.datastore.gorm.GormStaticApi$_methodMissing_closure2.doCall(GormStaticApi.groovy:101)
        at au.org.ala.images.SettingService.getOrCreateSetting(SettingService.groovy:109)
        at au.org.ala.images.SettingService.getSettingFromStack(SettingService.groovy:85)
        at au.org.ala.images.SettingService.getBoolSetting(SettingService.groovy:93)
        at au.org.ala.images.SettingService.getBackgroundTasksEnabled(SettingService.groovy:25)
        at au.org.ala.images.ProcessBackgroundTasksJob.execute(ProcessBackgroundTasksJob.groovy:17)
        at grails.plugins.quartz.GrailsJobFactory$GrailsJob.execute(GrailsJobFactory.java:104)
        ... 2 more
Caused by: org.hibernate.exception.JDBCConnectionException: could not inspect JDBC autocommit mode
        ... 9 more
Caused by: org.postgresql.util.PSQLException: This connection has been closed.
        at org.postgresql.jdbc2.AbstractJdbc2Connection.checkClosed(AbstractJdbc2Connection.java:820)
        at org.postgresql.jdbc2.AbstractJdbc2Connection.getAutoCommit(AbstractJdbc2Connection.java:781)
        ... 9 more
2018-06-05 06:43:46,648 [quartzScheduler_Worker-3] ERROR util.JDBCExceptionReporter  - This connection has been closed.
2018-06-05 06:43:46,649 [quartzScheduler_Worker-3] ERROR util.JDBCExceptionReporter  - This connection has been closed.

A suggestion from Simon is that the code from the following section may be useful to make phylolink more resilient to database interruptions:

https://github.com/AtlasOfLivingAustralia/volunteer-portal/blob/develop/grails-app/conf/application.yml#L303

A possible location for the local changes required may be in:

https://github.com/AtlasOfLivingAustralia/image-service/blob/master/grails-app/conf/DataSource.groovy#L37

I am are planning to store user's rating of images as user metadata. It will use two metadata named like and dislike. Each metadata will have a list of user ids.
In scenario where a user first liked an image and then subsequently update the image with a disliked, the logic breaks as dislike and like metadata will have the user's id. The is because the code does not permit empty values to be updated for a metadata. Permitting empty value will solved this issue.

AtlasOfLivingAustralia/data-management#219

The workflow for navigating to linked occurrence records is not intuitive and some users are finding it easier to use the support button to report misidentified species from images.ala.org.au, which is not ideal. In addition to improving the navigation (issue #43), it could also be useful to allow users to directly add flags to occurrence records from within image-service.

Refs tickets 25350 and 25351 in helpdesk

Some admin commands are being called by crawler bots, which implies that they don't have authentication or apikeys restricting access. In this case, the target is /ws/scheduleArtifactGeneration

==> /var/log/tomcat7/catalina.out <==
Index Image 229250751: 2 ms
Index Image 229250751: 1 ms
2018-11-02 10:59:39,223 [http-bio-8080-exec-17549] INFO  images.LogService  - Username: N/A IP: 150.229.66.12 Session: F0ACE69B5C8C22296F007E5E6F0D3C75 UA: Mozilla/5.0 (Linux; Android 6.0.1; Nexus 5X Build/MMB29P) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/41.0.2272.96 Mobile Safari/537.36 (compatible; Googlebot/2.1; +http://www.google.com/bot.html) URI: /grails/webService/scheduleArtifactGeneration.dispatch

==> /var/log/apache2/images.ala.org.au.ssl_access.log <==
66.249.79.252 - - [02/Nov/2018:10:59:39 +1100] "GET /ws/scheduleArtifactGeneration/5b6de1d6-7f86-4080-9f2f-b924cc2adc87 HTTP/1.1" 200 6419 "-" "Mozilla/5.0 (Linux; Android 6.0.1; Nexus 5X Build/MMB29P) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/41.0.2272.96 Mobile Safari/537.36 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)"

Something like:

• /image/{id}/original
• /image/{id}/thumbnail
• /image/{id}/thumbnail/square
• /image/{id}/thumbnail/large
  /cc @djtfmartin @m-r-c

I see in the biocache-store logs that loading an image whether or not image already exists (which is usually the case) is fairly slow.
One solution that I guess we can have is to create an API that accepts multiple image and return a response showing that which images are already stored in the media store.

This way can reduce the round trip time for calling the individual image loading API for the whole batch.

@ansell suggested we could reduce disk usage by removing the various thumbnail copies that have only a different coloured background (grey, white & black) and provide just transparent colour (PNG) & black (jpg).

Via Google Analytics - need to trigger GA API manually for images embedded in other pages(?)

Might want to also trigger a downloads stat event against the DP/DR/CO code but there will be no other metadata (email, reason - can be canned).

JSONP results should always have the content type text/javascript, just as JSON results should always have the content type application/json. If the correct content types are sent, then browsers do not have to rely on possibly guessing what the content type should be, which reduces one set of bugs that users of different browsers could experience.

An example of a JSONP URL that is currently returning Content-Type: text/html, causing bugs when X-Content-Type-Options is used (or a browser fails to guess what the document really is):

https://images-beta.ala.org.au/ws/getImageInfo/be52e264-1f2b-4fd3-ba30-c6479560d224?callback=jQuery112409585056373834636_1561348582021&_=1561348582022

The code for rendering results in these cases looks like it might be the following:

https://github.com/AtlasOfLivingAustralia/image-service/blob/grails3/grails-app/controllers/au/org/ala/images/WebServiceController.groovy#L347-L361

Grails 3 issue.

https://images-beta.ala.org.au/album/index

• crumb trail display is not formatted correctly (appears as numbered list, not a crumb trail) and is not integrated into the (new) existing crumb trail (above it).
• "New album" button not working (see below)
• "Upload" button redirects to Images home page with alert "Could not find image with id stagedImages!"
• Confusing link under albums section with text "Desktop version", does nothing when clicked (remove this)?

Click button "New album".
Expected output: a modal dialog appears:

Actual output: nothing happens (no error or JS error).

There are lots of images that have a broken images. E.g. http://images.ala.org.au/image/details/66140333 - in this case it appears the Flickr URL is no longer active and an error message has been cached as the image: http://images.ala.org.au/store/e/9/1/0/2d944025-4521-421b-92a5-ead7fb04019e/original

We need to identify all broken images and then remove them or fix them. Advanced search does not allow this (that I could find).

I tested Biocollect with the new image server and most of the uploads worked as expected. But one image got cropped and a black background was added. The original image used is attached.

Actual Biocollect record
Uploaded image

Attached image -

On the advanced search page, _Select all matching images_ doesn't work.

A reference to the open source apache imaging commons ImageMetadata was replaced in this otherwise unrelated commit in September 2018 to a proprietary Oracle/Sun JDK API:

60093ef#diff-f7b6547f437aa9b4c1c950736acdbe43R7

This causes it to break when used with OpenJDK that doesn't include this proprietary class.

Not sure if its the redirect or the resulting https URL but images uploaded from sightings.ala.org.au started failing when we added an Apache permanent redirect for / to https://sightings.ala.org.au. Images records were created but the images were all broken (404). Removing the redirect fixed it.

Ex: https://bioatlas.se/images/image/details/250408
in the title

404 returned from /ws/updateMetadata despite success. Response needs to send a 201 and a JSON body indicating success.

As reported by Scott Loarie 2nd Oct.

The /image/proxyImageThumbnail is used by external apps (BIE, Biocache) to load thumbnail version of images. This service shouldnt require any CAS authentication.

In the case of non-images being stored in images, the proxyImageThumbnail service currently returns HTTP 404 instead of a replacement generic or static icon:

https://images.ala.org.au/image/proxyImageThumbnail?imageId=c14c7ecb-849e-491c-82c3-8b6424b25b0f

It would improve browsing using biocache-hub if there was a valid image returned for thumbnail calls for sounds and other media types we are storing in the images service.

On home page, add a browse by set of links (facets) on left hand side:

life form (species group)
species subgroup
file type (image, sound, video)
license type

When remotely uploading images via http://images.ala.org.au/ws/uploadImage, top-level metadata (attributes of the Image class), as opposed to the Image.metaData (arbitrary data).

In particular there are two fields for Image that should be set:

• copyright
• attribution

if those fields are present in the params.metadata POSTed data.

Currently these values are going through to the "System" metadata but do not appear on the output of http://images.ala.org.au/ws/getImageInfo/{imageId} (unless the param includeMetadata=true is added.

https://images.ala.org.au/search/index

The user first clicks the "add criteria button and then selects a criteria field from the drop-down list. After selecting an option, the screen displays a "Loading..." message that never resolves. Looking at the console there is an error:

Mixed Content: The page at 'https://images.ala.org.au/search/index' was loaded over HTTPS, but requested an insecure XMLHttpRequest endpoint 'http://images.ala.org.au/search/imageFieldCriteriaFragment?searchCriteriaDefinitionId=3032104&format='. This request has been blocked; the content must be served over HTTPS.

Fix is to change all "http:" references in JS to use either // or https: - ideally making this a configurable option so it will work under both HTTPS and HTTP.

When I re-ingest a dwca with multimedia extension, It doesn't update the metadata information on the images.

What's the work around this issue?

Suggest detecting the duplicate image hash and occurrence id during image creation and updating image metadata instead of creating a new image.

migrated from: https://code.google.com/p/ala/issues/detail?id=729
date: Wed Jul 2 18:29:56 2014
author: moyesyside

The image service pages are using the old style breadcrumbs:

Old skool:

https://www.dropbox.com/s/h3gkre4supz6nus/Screenshot%202014-07-03%2011.28.52.png

new skool:

https://www.dropbox.com/s/d3luljmclb93qv1/Screenshot%202014-07-03%2011.28.46.png

How is it possible to have access to admin of images module without the ALA-CAS?

Comes from Freshdesk ticket https://support.ehelp.edu.au/a/tickets/27235

The major issue of course was the insertion of bird data into a spider space.
A third problem I will mention again is the urgent need to eliminate identical pictures. It is extremely frustrating when I have to spend many hours wading through thousands of photos which appear in random order and many of them are identical duplicates and some of them appear up to half a dozen times.

Need to be able to run batch updates to image metadata particularly licenses:

e.g. set license for all or a subset of images from a data resource

• the set default image license in the collectory seems to be over riding the information provided in a Darwin Core archive and it doesn't have an option to set the default back to empty.

Specific example is dr660 (ANFC images) - I'm trying to update the licensing for a subset of the images to "All rights reserved" but the images still have CC-BY_NC on them, see dr660 metadata for file and current settings.

We found a hard coded URL reference in "plugins/images-client-plugin-0.8/js/ala-image-viewer.js":

    var imageServiceBaseUrl = "https://images.ala.org.au";

This broke some js interactions in our runtime (can't recall where right now). We are patching this post-install in our setup with sed, maybe ansible playbooks should do the same.

Grails 3 branch.

Current prod version allows the user to select images by clicking the "tick box" next to each image or using the drop-down "cog" icon to either "select all on page" or "select all matching images". Once one or more images are selected, the text in the "crumb trail portion of the page (right side)" shows:

This functionality appears to be lost in the new version, as neither the "tick boxes" nor the "X selected images" tool (screenshot above) are displayed.

Therefore a user cannot create an album, which is a requirement for exporting images as CSV file.

Fixes:

• "cog" icon missing on drop-down tool on right side of screen (after doing a search)
• images from search are missing "tick box"
• tool to allow an album to be created or to show existing albums is missing, so re-instate

Some images taken with smart phones, etc come through with the wrong orientation in the generated thumbnail, etc. The original version of the image is OK but all other generated images are usually 90° off.

I have a feeling this is a limitation of the image generation library used???

Suggestion from @sadeghim is to store a checksum on each image and then to warn (flag) or error when a new image is uploaded that has the same checksum.

Would also be good to have a find duplicate images for existing images. How to handle historical duplicate images is another matter that would require further investigation.

caption and description are currently added to the biocache record but need to be stored with the image - e.g. for APII images

Upgrades of the following:

• Bootstrap3
• Grails 3
• Elastic search 7

The image-service API is not currently described on api.ala.org.au, making it difficult to know what functions are available.

findImagesByMetadata from version 0.5 onwards have stopped working.
Im not clear why this is the case. Heres the code for query building:

        def filter = FilterBuilders.orFilter()
        values.each { value ->
            // Metadata keys are lowercased when indexed
            filter.add(FilterBuilders.termFilter(key.toLowerCase(), value))
        }

An example document (note the nested element 'metadata')

{
    "took": 3,
    "timed_out": false,
    "_shards": {
        "total": 5,
        "successful": 5,
        "failed": 0
    },
    "hits": {
        "total": 1,
        "max_score": 3.7037592,
        "hits": [
            {
                "_index": "images",
                "_type": "image",
                "_id": "48",
                "_score": 3.7037592,
                "_source": {
                    "imageIdentifier": "84350f65-19e7-4ccf-a2b6-807525d4e330",
                    "contentMD5Hash": "cef944bf2d88d36fd6ef0a9568d5a431",
                    "contentSHA1Hash": "28de413233afaac734164e01aafb18c3006b6cc4",
                    "mimeType": "image/jpeg",
                    "originalFilename": "dr4||7bfe0532-0df3-41c1-9707-97a15019763f||smallRaw.jpg",
                    "extension": "jpg",
                    "dateUploaded": "2015-08-10T10:00:04Z",
                    "dateTaken": "2015-08-10T10:00:04Z",
                    "fileSize": 22363,
                    "height": 209,
                    "width": 314,
                    "zoomLevels": 0,
                    "dataResourceUid": "dr4",
                    "creator": "Dave Martin",
                    "title": "My image title",
                    "description": "My image description",
                    "rights": "All rights reserved",
                    "rightsHolder": "Flora of Australia",
                    "license": "Creative Commons Attribution-Noncommercial-Share Alike",
                    "thumbHeight": 0,
                    "thumbWidth": 0,
                    "harvestable": false,
                    "metadata": {
                        "component 3": "Cr component: Quantization table 1, Sampling factors 1 horiz/1 vert",
                        "number of components": "3",
                        "x resolution": "1 dot",
                        "resolution units": "none",
                        "image height": "209 pixels",
                        "data precision": "8 bits",
                        "compression type": "Baseline",
                        "image width": "314 pixels",
                        "version": "1.2",
                        "y resolution": "1 dot",
                        "component 1": "Y component: Quantization table 0, Sampling factors 2 horiz/2 vert",
                        "component 2": "Cb component: Quantization table 1, Sampling factors 1 horiz/1 vert",
                        "fulloriginalurl": "http://bie.ala.org.au/repo/1013/207/2070645/smallRaw.jpg",
                        "originalfilename": "smallRaw.jpg",
                        "occurrenceid": "7bfe0532-0df3-41c1-9707-97a15019763f",
                        "identifier": "http://bie.ala.org.au/repo/1013/207/2070645/smallRaw.jpg"
                    }
                }
            }
        ]
    }
}

cc @m-r-c @aruizca - if you guys have insights as to what might have changed in elastic search I'd been keen to know. I couldnt spot anything obvious in release notes.

Hi all,

Just a quick follow up on the automatic image tagging from "Melbourne Science Hackfest" event.

After a bit of playing around, found following third party backend service that has been used by the app (http://115.146.94.208/) to get the image context.
Its free to use for the first 5000 request per month.

Service Provider: https://developer.clarifai.com
Pricing for the service: https://developer.clarifai.com/pricing/

Also, there is one open source framework which will also allows us to get the image context.
Demo: http://demo.caffe.berkeleyvision.org
Website link: http://caffe.berkeleyvision.org

Cheers
Sathish

We get this SQL grammar exception:

2019-02-25 12:51:00,516 [imagenes.gbif.es-startStop-1] ERROR util.JDBCExceptionReporter  - ERROR: column am.amcanorder does not exist
  Position: 427
2019-02-25 12:51:00,519 [imagenes.gbif.es-startStop-1] ERROR hbm2ddl.SchemaUpdate  - could not complete schema update
org.hibernate.exception.SQLGrammarException: could not get table metadata: album
        at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
        at java.util.concurrent.FutureTask.run(FutureTask.java:266)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
        at java.lang.Thread.run(Thread.java:748)
Caused by: org.postgresql.util.PSQLException: ERROR: column am.amcanorder does not exist
  Position: 427
        at org.postgresql.core.v3.QueryExecutorImpl.receiveErrorResponse(QueryExecutorImpl.java:2270)
        at org.postgresql.core.v3.QueryExecutorImpl.processResults(QueryExecutorImpl.java:1998)
        at org.postgresql.core.v3.QueryExecutorImpl.execute(QueryExecutorImpl.java:255)
        at org.postgresql.jdbc2.AbstractJdbc2Statement.execute(AbstractJdbc2Statement.java:570)
        at org.postgresql.jdbc2.AbstractJdbc2Statement.executeWithFlags(AbstractJdbc2Statement.java:406)
        at org.postgresql.jdbc2.AbstractJdbc2Statement.executeQuery(AbstractJdbc2Statement.java:286)
        at org.postgresql.jdbc2.AbstractJdbc2DatabaseMetaData.getIndexInfo(AbstractJdbc2DatabaseMetaData.java:4234)

This is a ubuntu xenial server with postgresql-9.6.

It seems that the posgresql driver should be updated.