bug: Management API is unreasonably slow

[Joonsey]

Management API is unreasonably slow

We have a project utilizing some of the features on the 'Management API'.

In particular we query '/organizations//users' very regularly. And this endpoint is very slow.

Expected behavior

I expect this endpoint to be faster, at least median lower than 360ms

How to reproduce?

curl request the management api + '/organizations//users'

Context

• [x] Self-hosted, Logto version = 1.9.1
  • [x] Container (Docker image)

Here is a dump of the grepped calls to this endpoint (this is obtained from the logged output of the container, meaning it is not accounting for dns resolution, tls / ssl etc. At least to my knowledge): dump.txt

some aggregates:

• count: 18129
• sum: 6704359ms
• avg: 369.8ms
• max: 3767ms

[charIeszhao]

Thanks for reporting. We'll take a look. What's the average response time for your other requests?

[Joonsey]

Off the top of my head they are generally faster. The oidc requests etc. are single digit latency, other management api requests ranging from 12-200ms from the top of my head.

Although i can double check for you.

[charIeszhao]

Thanks

[Joonsey]

all_dump.txt

Here is a dump of all the different responses to our logto instance, of course this includes the management api requests and also oidc request.

aggregates

• avg: 225.8ms
• sum: 7427582ms
• count: 32901

To elaborate;

if we remove the previous aggregates which are all the calls to 'organizations//users' we get these results:

• avg: 49.0
• sum: 723223ms
• count: 14772

To me it is clear that the 'organizations//users' is significantly slower than the rest of our generally used endpoints