Bug #4350
closedBug #4252: Slow queries for list precalculation (heavy parallel sequential scan 8-20 sec per each 10 mins)
Enum table not populated for cesnet_inspectionerrors
100%
Description
The table enum_cesnet_inspectionerrors that should contain a list of unique values from the cesnet_inspectionerrors array is empty. This is probably due to an omission during the implementation of the new enumeration gathering system (and even more so during the testing).
A test query shows that there are multiple possible existing values.
This was confirmed on all hub, alt, dev.
The impact is probably none as there does not seem to be any user of this enumeration (a probable reason it slipped during testing). Does this enum need to exist at all? Or should the inspection errors editbox (only visible in Admin mode) be in fact a drop-down selector?
Related issues
Updated by Jan Mach over 5 years ago
- Related to Task #4228: Hawat: Make use of Flask-Cache plugin added
Updated by Jan Mach over 5 years ago
- Status changed from In Progress to Feedback
- Assignee changed from Jan Mach to Radko Krkoš
- % Done changed from 0 to 100
Fix deployed to mentat-alt for testing, but at this point the database is stale and the mentat-precache.py module is unable to finish. The event search page returns a 500 error because of the database unavailability.
Updated by Jan Mach over 5 years ago
Database works now, the long running operations are now finished.
Updated by Radko Krkoš over 5 years ago
- Assignee changed from Radko Krkoš to Jan Mach
I seem not to understand the problem here. Please explain. Was it the 1000s run time of the first data load?
Updated by Jan Mach over 5 years ago
- Assignee changed from Jan Mach to Radko Krkoš
Radko Krkoš wrote:
I seem not to understand the problem here. Please explain. Was it the 1000s run time of the first data load?
Yes, there were some heavy database operations and the first run time was really high. I just wanted to log somewhere, that because the first run was taking so long the web interface was no working due to the missing cache file. I also logged that for future reference and for your convenience. I think that unless there was some serious problem with the database at that time we can disregard this note.
In my opinion this issue can be closed, but I am leaving that up to the author of the bug request.
Updated by Radko Krkoš over 5 years ago
- Status changed from Feedback to Closed
- Target version set to 2.2
Jan Mach wrote:
Radko Krkoš wrote:
I seem not to understand the problem here. Please explain. Was it the 1000s run time of the first data load?
Yes, there were some heavy database operations and the first run time was really high. I just wanted to log somewhere, that because the first run was taking so long the web interface was no working due to the missing cache file. I also logged that for future reference and for your convenience. I think that unless there was some serious problem with the database at that time we can disregard this note.
Yes, I see. Thanks for explanation. Such run time is unfortunately to be expected as the common optimization of parallel scan is disabled in PostgreSQL 10 in data modification queries (in 11 it is enabled for simple queries such as this).
In my opinion this issue can be closed, but I am leaving that up to the author of the bug request.
Agreed, closing, I am happy with the result.