Page MenuHomeSoftware Heritage

[swh-search] Investigate long search queries response time
Closed, MigratedEdits Locked

Description

Most of the search queries take around 3-5 seconds which isn't good for UX. We should do some profiling tests, find out the bottlenecks, and fix them.

P1364 contains a list of queries for which we should run Kibana profiling tests to understand if the problem originates from elasticsearch itself. If it doesn't, we need to investigate other parts of the stack.

Related Objects

Event Timeline

KShivendu created this object in space S1 Public.
KShivendu added a subscriber: vsellier.
vsellier triaged this task as Normal priority.May 17 2022, 8:54 AM
vsellier removed a subscriber: vsellier.

Here are the results of the queries.
You can directly paste the json in the search profiler to see the result.
(Be careful some are quite huge)

  • visited = true

  • (((visited = true and visits > 0)))

  • origin : django and metadata : "web"

-origin : django

  • metadata : "web"

  • origin : django or metadata : "web"

  • visits != 5

  • visit_type = [git,"pypi"]

vsellier renamed this task from Improve search performance to [swh-search] Investigate long search queries response time.May 17 2022, 5:38 PM
vsellier closed this task as Resolved.
vsellier reopened this task as Work in Progress.
vsellier claimed this task.
vsellier edited projects, added System administration; removed System administrators.
vsellier moved this task from Backlog to in-progress on the System administration board.

@KShivendu Any news regarding these profiling ?

Should we keep this ticket open?

Please keep it open since the search isn't fast enough at the moment.