Redmine

Ok great. When i get back to work tmr i'll post the debug logs as requested.

In the meantime can you please try 50 projects each with a subproject that has itself 10 subprojects as this would more closely match our setup
I have also edited redmine file i think it was query.rb to remove entire project list from issues filter section and get a huge speed increase for this page

Strangely all these speed increases are only seen for anyone logged in as an adminustrator. Normal users still experience 5secs, admins 500ms.

Am using latest bitnami redmine stack, but have seen this problem for years but never found the problem

Gut feeling its related to the number of levels of nedted suprojects. No ides why admin would see faster times than normal user

Dell Green wrote:

In the meantime can you please try 50 projects each with a subproject that has itself 10 subprojects as this would more closely match our setup

I've tried this exact setup with current trunk and 2.6-stable, response times are not much different.

Started GET "/" for 127.0.0.1 at 2015-02-15 09:07:26 +0100
Processing by WelcomeController#index as HTML
  Current user: xxxx (id=xxxx)
  Rendered news/_news.html.erb (3.0ms)
  Rendered welcome/index.html.erb within layouts/base (7.0ms)
Completed 200 OK in 422ms (Views: 124.0ms | ActiveRecord: 24.0ms)

Hmm, ok thanks. I'll post the debug logs tmr, maybe this will shed some light on the problem.

It must be my data is some way as i have installed the data on fresh installs of bitnami redmine on mac and linux and the problem is the same.

Please let me know if you require any further information.

We are experiencing the exact same issue:

1. long project list (around 300) – we have just 3 sub-projects.
2. Redmine v3.3.0
3. load time for normal user (same number of projects, roles, permissions and same page): 5 sec (avg)
4. load time for admin user: 1.5 sec (avg)

This was tested in a staging machine on Bitnami Redmine with Apache, Passenger and MySQL. Single request mode.
Is there something else we can do to gather more information about this issue?

After other research on the subject we found what follow:

1. removing the quick jump bar does not solve the problem at all;
2. users participating in few (10-20) projects have lighting fast responses (in 8-500 ms range);

There should be something related to the number of projects in connection with the membership and permissions check; the point is that it happens even inside a project and not only in the projects list, where there's no reason (apparently) to check permissions on each request. This problem makes Redmine not usable with more then a few project (actually we have just 350).

So, anyone has some suggestions on how to delve deeper in this issue?
Thanks!

+1

I have ~1000 projects and ~100 users and have same problem.

The performance is good (loading pages between 0,3 and 1,0 second) for admin account, but when access to the same pages with other users (with limited permissions), the performance is bad (loading pages between 5 and 7 seconds)

In debug mode, I can see that the SQL requests are very big with users permissions on hundred of projects (a lot of project_id).

JL, a french user.

+1

When member of a group that's in 600 projects (actually, member of one top project with 600 sub-projects with inherit members enabled), the issues page takes 3-4 sec to show. When not a member it takes a few 100 ms. (Non-members can view issues too).

RM 3.1.1-stable
Ruby 2.1.7-p400 [x64-mingw32]
Rails 4.2.7.1
DB: SQL Server 2014

+1

Environment:
Redmine version: 3.3.3.stable
Ruby version: 2.1.10-p492 (2016-04-01) [x86_64-linux]
Rails version: 4.2.7.1
Environment: production
Database adapter: Mysql2
SCM:
Filesystem
Redmine plugins:
no plugin installed

As an update. Previously i was running our redmine instance on a mac mini core i7 with mac os, recently i moved it over to an HPE Proliant dl380 g6 runnng linux and the exact same problem exists. In both cases whenever anyone loads a page i see a ruby process for each user running at 100% for 5 to 10 seconds while the page is retrieved for each user. So its not platform or resource related.

Any tips for reducing the query time would be greatly appreciated.

About:

As an update. Previously i was running our redmine instance on a mac mini core i7 with mac os, recently i moved it over to an HPE Proliant dl380 g6 runnng linux and the exact same problem exists. In both cases whenever anyone loads a page i see a ruby process for each user running at 100% for 5 to 10 seconds while the page is retrieved for each user. So its not platform or resource related.

I can confirm it. We increased the vCPUs of our server (2 to 4 Intel Xeon E5-2640 2.60GHz cores) and the exact same problem exists. This is clearly an application issue.

It looks like the issue was fixed in 3.4.0.

@Dell Green, Luca Lenardi, Jean Louis & Anders Thomsen,

As posted by Renê Barbosa, there are several performance improvements included in 3.4.0 that might fix your issues too. Could you all please test with Redmine 3.4.0 and report back your feedback in this issue? Thanks in advance.

Yes, performance is much better now. Still not as fast as with admin users though.

yes, many thanks