https://datahubproject.io logo
Join Slack
Powered by
Hi everyone! I recently upgraded to v0.9.6.1 (fro...
# troubleshoot
b
Hi everyone! I recently upgraded to v0.9.6.1 (from v0.9.2) and am encountering a strange bug. When I am on the explore page (
Explore All
button from the homepage), if I try to use the advanced filter, some of the options like filtering on tags does not work. It seems that the API call which is supposed to populate the dropdown is not returning anything. However, it does seem like the 
basic
filtering works (the issue is related to the API call which is supposed to populate the dropdown menus). Does anyone have an idea of what is happening and/or have an idea of how to fix it? Here is a step by step set of instructions on how to reproduce the issue (start on the homepage of DataHub): 1. Select 
Explore All
2. Under the 'Filter' column, select 
Advanced
3. Select 
Add Filter
4. Select 
Tag
5. Search for a tag which exists in your datahub instance (in my case, I always get 'no data')
e
Hi @bumpy-pharmacist-66525 are you seeing any Error Logs?
It seems to be working okay for me on the demo website
b
Hi @echoing-airport-49548, I am not seeing any errors in the logs. I took a look at the GraphQL query which is being sent so that the front-end can pre-populate the drop down, however, it seems to always return an empty list of results. One thing I did notice is that if I search for a specific, existing tag, and click the 
Create X
option in the dropdown, the tag in the dropdown menu will then properly link to the already existing tag called 
X
. After this, you can search up the tag and see it in the dropdown of the advanced filter. It seems I am not the only one experiencing this issue, there is another post which describes the same bug: (https://datahubspace.slack.com/archives/C029A3M079U/p1675971293250959)
a
CC: @big-carpet-38439
b
Just to add more information, the dropdown does not populate for the following options within the advanced filter: 
Platform
Tag
Column Tag
(The bug might impact more of the options, but these 3 are the ones that I have tested and can confirm that there is an issue with them)
I went through all of the advanced filter options and wrote down which GraphQL operation they each use. Here is a list of my findings:
Copy code
Entity Type: N/A
Platform: getSearchResults
Owner: getSearchResultsForMultiple
Tag: getSearchResults
Glossary Term: getRootGlossaryTerms
Domain: getSearchResultsForMultiple
Column Tag: getSearchResults
Column Glossary Term: getRootGlossaryTerms
Column Name: N/A
Column Description: N/A
Description: N/A
Container: getSearchResults
Soft Deleted: N/A
Sub Type: N/A
Environment: N/A
It seems that any of the options which use the 
getSearchResults
operation to populate the dropdown menu are the ones that are currently having issues. E.g., Platform, Tag, Column Tag, Container (I did not test this one, but I presume it's also not working)
I also just noticed that wherever you can assign tags to something (e.g., datasets, columns on a dataset), the dropdown menu is also not populating there. So it seems the issue is not isolated to the advanced filter.
b
This appears to me that the search indexing is not working properly... Did the only happen after upgrading?
I have not being able to reproduce
So once you create a tag, suddenly everything starts to work?
Are other parts of the app (Domains, Users, Groups) loading properly
b
@big-carpet-38439 that is correct, this behavior only began happening after upgrading to v0.9.6.1. That is also correct, if when I am searching for a tag (e.g., 
tag1
) and I click the create button, after that is done, 
tag1
will 'link' to the already existing 
tag1
tag. After this, 
Tag1
starts to appear in the dropdown menu for tags. However, I would need to manually repeat this process for any pre-existing tag (before the upgrade), which is less than ideal. In terms of other parts of the app loading correctly, right after the upgrade the domains were not showing up in the 
Govern -> Domains
page. In order to get them to show up I added a 'test' domain and the list all of a sudden got updated (with the actual domains), and it has been fine ever since. I am not too sure if this is related to the search indexing issue which is happening with tags, platform, etc. In terms of Users and Groups, they seem to be working as expected. I did follow this guide: https://datahubproject.io/docs/how/restore-indices/ and the Platform dropdown started to populate with results. However, the tag dropdown is still not populating correctly. You mentioned that the search indexing might not be working properly, is there a way to re-index it (other than the guide above) so that the tags start to populate in the dropdown(s)?
@big-carpet-38439 I was looking through the logs of the 
datahub-upgrade
container, and I found this (see attached screenshot). I am not too sure if this is what is causing the issue, but I thought I would post it just in case.
b
So this all actually looks good
Do you see any errors on the 
datahub-gms
pod?
Anything related to elastic search?
b
I don't see any errors related to elastic search on the 
datahub-gms
pod
Hi @big-carpet-38439, after looking at the MySQL db, it seems that the tags were deleted during the upgrade. This would explain why they are not appearing in the dropdown menu for filters, as they simply don't exist anymore. The only thing that seems to remain of them are any references (to the tag) that are stored in the dataset entities themselves. I am still unsure about what went wrong in the upgrade which would have caused this behavior. The workaround we have for the tag issue is to manually create each of the tags which used to exist, and they seem to work fine after that. For the platform issue, once we ran the re-indexing cron job the list started to populate. For the final issue of domains not displaying, once we added a test domain, all domains started showing up again (including the newly-created test domain).
@big-carpet-38439 To give an update in regards to this bug. I tried to reproduce the issue by creating a new instance of DataHub running v0.9.2. Once it was running, I ingested some datasets into it and added tags to them. I also created some dummy domains as we are also having issues with domains not appearing in the domains page (Govern -> Domains) after the upgrade. After this was done, I completed the upgrade process as we had done previously (to go from v0.9.2 -> v0.9.6.1) and looked at the results. It seems that the tags were not deleted during this upgrade, and were now available in the filtering dropdown. The platform dropdown also seems to be working. However, the domains page once again did not have the domains which were there before the upgrade. To allow the domains to show up, I simply needed to add a test domain, and once I did, the list populated with the preexisting domains. If you would like, I could create an issue for this on the DataHub GitHub repo. Although the platform dropdown 'worked' after the upgrade, I did notice it behaved strangely. For example, when you first opened up the advanced filtering - platform modal, it displayed 
no data
. Only once you entered in 2 or 3 characters of one of your platforms did it actually give you results in the dropdown. Another behavior is that when you backspaced all the way (i.e., no characters in the search bar), it shows you platforms which are not on your DataHub instance. This is another thing I can create an issue for if you would like.
b
Thanks for the detailed update!
So I have no idea why your Tags would have been gone from DataHub. Have you issued any DataHub deletes recently?
The Platform issue is definitely a known issue and it's on my backlog to pre-populate the dropdown with options 🙂
The Domains one is definitely a bug. It should absolutely be loading the entire list
cc @brainy-tent-14503
b
In terms of issuing any deletes recently, I haven't run any in a month or two. However, I did run them while we had v0.9.2 deployed.
Open in Slack
PreviousNext
Powered by