Bug 844948
| Summary: | The quota didn't change after move app from medium district to small district | ||||||
|---|---|---|---|---|---|---|---|
| Product: | OKD | Reporter: | Rony Gong 🔥 <qgong> | ||||
| Component: | Containers | Assignee: | Rob Millner <rmillner> | ||||
| Status: | CLOSED CURRENTRELEASE | QA Contact: | libra bugs <libra-bugs> | ||||
| Severity: | medium | Docs Contact: | |||||
| Priority: | medium | ||||||
| Version: | 2.x | CC: | cgole, dmcphers, mfisher | ||||
| Target Milestone: | --- | Keywords: | Triaged | ||||
| Target Release: | --- | ||||||
| Hardware: | Unspecified | ||||||
| OS: | Unspecified | ||||||
| Whiteboard: | |||||||
| Fixed In Version: | devenv_1933 | Doc Type: | Bug Fix | ||||
| Doc Text: | Story Points: | --- | |||||
| Clone Of: | Environment: | ||||||
| Last Closed: | 2012-08-07 20:43:03 UTC | Type: | Bug | ||||
| Regression: | --- | Mount Type: | --- | ||||
| Documentation: | --- | CRM: | |||||
| Verified Versions: | Category: | --- | |||||
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||
| Cloudforms Team: | --- | Target Upstream Version: | |||||
| Embargoed: | |||||||
| Attachments: |
|
||||||
Why is the inodes limit different for small vs medium (or any gear size using 1G for that matter)? From what we are selling we say they are both 1G. I would expect changing gear sizes wouldn't give me more files I can add within the same space. The issue being that going from a medium to small would mean my same data might not fit (once we allow users to make these kinds of choices). [rmillner@splat stickshift]$ grep quota_files * resource_limits.conf:quota_files=40000 resource_limits.conf.c9:quota_files=40000 resource_limits.conf.exlarge:quota_files=400000 resource_limits.conf.high_density:quota_files=40000 resource_limits.conf.jumbo:quota_files=4000000 resource_limits.conf.large:quota_files=40000 resource_limits.conf.medium:quota_files=468114 resource_limits.conf.micro:quota_files=20000 resource_limits.conf.small:quota_files=40000 It seems accidental that the medium quota is different than either the large or the small. When you migrate from medium to small; should the quota retain medium's settings or should it force quota small? The former means we risk over-allocating disk on the small ex nodes. The latter means we either risk a failure if the app's disk usage is beyond the limit or have to build in additional checks. For this sprint; I'm going to verify that moving the quota is intentional behaviour. If so, then this will move to FutureFeature to determine what the correct behaviour should be. From the logs and code; it seems as though we're intentionally setting the new quota based on the old on. Moving to FutureFeature to discuss if that's the correct semantics. That is the correct semantics. Once they are given more space we can't just take it away. But the issue is we set a different inodes limit for small vs medium gears. I think that was done because the medium gears had less apps. But I believe that's the wrong logic. We should keep the same 1G standard across all gear sizes. Agreed - and its a quick fix to set medium's quotas to the same as small and large. Pull request: https://github.com/openshift/li/pull/161 Taking off future Feature since its being dealt with in this sprint. Pull request accepted. Verified - all node profiles now have same default quota limit i.e. 40000 To test - 1. Created 2 districts, one small one medium 2. Added a node with a small profile to the small distrct and a medium profile node to the large district 3. Create a app. 4. login to the app and type quota. Make sure the files limit is 40000 5. Check which district the app is in and move the app to the other district 6. login to the app and type quota. Make sure the files limit is 40000 |
Created attachment 601705 [details] development.log Description of problem: The quota didn't change after move app from medium district to small district Version-Release number of selected component (if applicable): devenv_1931 How reproducible: always Steps to Reproduce: 1.Setup multi node env, and create 2 districts, small and medium 2.Create a medium app, eg php 3.Move this php to small node across district. 4.Check the quota of this app Actual results: Still large for that 468114 [qphp-qgong1.dev.rhcloud.com ~]\> quota Disk quotas for user 1263e78e26a343ab9050554ea167d321 (uid 1002): Filesystem blocks quota limit grace files quota limit grace /dev/xvde2 620 0 1048576 159 0 468114 Expected results: The files limit should be 40000, small Additional info: I find if move app from small to medium district, the quota could change to meidum.