-
Bug
-
Resolution: Fixed
-
Low
-
None
Get a sorted list of searchable fields is more than 160 times slower in JIRA 6 than in JIRA 5.2. This comes from the fact that translatable field names was introduced in JIRA 6 which means that JIRA has to get a property set for each field where the translated name is stored for that field when it get the field's name now.
These property sets should be cached but it looks like if there is a high number of fields most of the caches are empty and have to be looked up from the DB from scratch which is taking all the time.
We need to introduce a more performant field name cache in JIRA.
Workaround
Upgrade to JIRA 6.0.1
- is related to
-
JRASERVER-32510 Very slow response on report configuration page with 10K crowd groups
-
- Gathering Impact
-
-
JDEV-29451 Failed to load
- is cloned from
-
JRADEV-21343 Failed to load
- mentioned in
-
Page Loading...
-
Page Loading...
-
Page Loading...
-
Page Loading...
-
Page Loading...
-
Page Loading...
-
Page Loading...
-
Page Loading...
-
Wiki Page Loading...
-
Wiki Page Loading...
-
Wiki Page Loading...
-
Wiki Page Loading...
-
Wiki Page Loading...
-
Wiki Page Loading...
-
Wiki Page Loading...
-
Wiki Page Loading...
-
Wiki Page Loading...
-
Wiki Page Loading...
-
Wiki Page Loading...
-
Wiki Page Loading...
-
Wiki Page Loading...
-
Wiki Page Loading...
-
Page Loading...
-
Wiki Page Loading...
Form Name |
---|
There were two problems involved:
1. The custom field translation lookup was supposed to be cached, but the cache wasn't working (the new problem introduced by the translations), which increased the time needed to resolve the name of a custom field.
2. The report configuration page had a problem that caused the sorted set to be recalculated once per field instead of reusing a single sorted set. This was an existing problem.
On the report configuration page, these two problems combined to make that page particularly slow. Both of them are addressed in 6.0.1/OD-15.