Bug 1371277 - etcd cache size default value set to 50K causing high memory usage
Summary: etcd cache size default value set to 50K causing high memory usage
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: OpenShift Container Platform
Classification: Red Hat
Component: RFE
Version: 3.3.0
Hardware: Unspecified
OS: Unspecified
medium
medium
Target Milestone: ---
: ---
Assignee: Timothy St. Clair
QA Contact: Chuan Yu
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2016-08-29 18:33 UTC by Vikas Laad
Modified: 2017-02-13 14:21 UTC (History)
7 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2017-01-25 22:11:16 UTC
Target Upstream Version:


Attachments (Terms of Use)

Description Vikas Laad 2016-08-29 18:33:15 UTC
Description of problem:
Please see bug https://bugzilla.redhat.com/show_bug.cgi?id=1323733 using this bug etcd cache size was made configurable so that high memory consumption can be avoided. Currently default value is still set to 50K which causes again same problem until its set to lower number in master-config.

Version-Release number of selected component (if applicable):
openshift v3.3.0.22
kubernetes v1.3.0+507d3a7
etcd 2.3.0+git


How reproducible:
Long running tests can reproduce this problem.

Actual results:
Default value set to 50K causes high memory consumption.

Expected results:
This value should be set to lower number by default, not sure what that number should be.

Additional info:

Comment 1 Andy Goldstein 2016-08-30 14:05:28 UTC
Not a release blocker given that you can set this value in the master config.

Comment 2 Seth Jennings 2016-09-08 14:31:22 UTC
Current PR:
https://github.com/openshift/origin/pull/10719

Comment 3 Derek Carr 2016-10-26 17:31:35 UTC
Not a release blocker, need to have platform team decide if origin wants to change how caches are handled.  We can iterate on Seth's PR, but its not really a bug.  Personally, I think this should be closed and we should have an RFE to handle this.  Moving this to RFE component.

Comment 7 Timothy St. Clair 2017-01-25 22:11:16 UTC
The original title of this BZ no longer applies, I will open an upstream issue on determining proper sizing of the deserialization cache, and link back to this issue.


Note You need to log in before you can comment on or make changes to this bug.