Describe the issue:
We were using OpenSearch quite a long.
Recently I’ve upgraded my test server to 2.10.0.
It turns out that new Discover is utterly unusable with typical laptop screen because it is only 5% of screen can be used to browse my search results.
Configuration:
Quite typical.
Question is if there is any chance that it will be possible to switch to old Discover after 2.11 release or should we switch to Elasticsearch ASAP?
@rlevitsky We’d definitely like to better understand your use case and concerns. The overall layout of the new discover app is intended to be very similar.
Do you have some screenshots you can share to help us better understand?
@ashwinpc Is there an official feedback post on the forum? I didn’t find it with a quick search.
@rlevitsky Can you share a recording of the issue you are facing? We did get complaints that because the histogram and query bar was fixed to the page, the remaining scroll area for the data table itself was much smaller and hard to use. I’ve fixed that issue and the whole page should be scrollable in 2.11. You can pull the latest version from our main branch and test the changes to see if they address your concerns
@rlevitsky Thanks for the screenshots. Yes the whole page now scrolls, however the side panel isnt collapsible just yet. That is something we are still working on.
I do not have a docker version with the changes just yet, but the release builds will be ready soon and you can test it out then
The interface is better now, but still has some deficiencies to be made up.
For example, in previous versions the column widths in a table were auto-sized based on their contents. For log data where you may have fields like timestamp, server name, severity, and message the appropriate widths for the fields other than the message are naturally narrow. The auto-sized widths would make the columns with short entries narrow and give the remainder of the screen width to the message field.
In the current version the time column is auto-sized but the remainder of the screen width is split equally between the rest of the columns. For some datasets this leaves a lot of empty whitespace between columns. This is a minor annoyance as the widths can be manually adjusted, but unfortunately there are events that trigger some type of refresh that resets the column widths back to default. This is a major annoyance.
When viewing surrounding documents, using the load newer/older document features trigger a column width reset.
I have not narrowed it completely down, but there are cases where actions such as adding/removing a filter, modifying the time range or refreshing the view will trigger a reset of the column widths on the standard discover view, but it doesn’t happen every time.
I did not see any relevant settings in 2.11 that would change this behavior. Is there anything on the roadmap concerning auto-sizing column widths or allowing the column widths to be saved or at least persist in a session without being reset back to default?
The limitation of 100 lines per page is very annoying. In the previous version it was possible to scroll down and load more lines and the total limitation was configurable.
In the previous version it was possible to expand multiple long lines (using the > on the left side), to compare contents of long lines. In the new version it’s only possible to “expand” a single line, by opening a side panel with the line/document details.
Also, as already mentioned before me, the ability to hide the left panel is really useful in laptop screens.