Bug 1290667 - Files not promoted in the intended cycle are getting promoted in the next cycle(spilling over)
Summary: Files not promoted in the intended cycle are getting promoted in the next cyc...
Keywords:
Status: CLOSED WONTFIX
Alias: None
Product: Red Hat Gluster Storage
Classification: Red Hat Storage
Component: tier
Version: rhgs-3.1
Hardware: Unspecified
OS: Unspecified
unspecified
urgent
Target Milestone: ---
: ---
Assignee: hari gowtham
QA Contact: Nag Pavan Chilakam
URL:
Whiteboard: tier-migration
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2015-12-11 05:38 UTC by Nag Pavan Chilakam
Modified: 2019-04-03 09:15 UTC (History)
2 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2018-02-06 17:42:55 UTC
Embargoed:


Attachments (Terms of Use)
summerised log is easy qe log which is crisp (59.42 KB, text/plain)
2015-12-11 05:38 UTC, Nag Pavan Chilakam
no flags Details
node#1 brick logs zod (75.33 KB, text/plain)
2015-12-14 08:37 UTC, Nag Pavan Chilakam
no flags Details
node#2 brick logs yarrow (75.47 KB, text/plain)
2015-12-14 08:37 UTC, Nag Pavan Chilakam
no flags Details
tier cli logs (7.18 KB, text/plain)
2015-12-14 08:38 UTC, Nag Pavan Chilakam
no flags Details

Description Nag Pavan Chilakam 2015-12-11 05:38:25 UTC
Created attachment 1104568 [details]
summerised log is easy qe log which is crisp

Description of problem:
======================
I have set watermarks low to 1%(~200MB) and hi as 5%(~1GB) with my hot tier being about 20GB(1x2 hot tier over (4+2)+2 ec vol)
With settings remaining same as above,I heated about 10files each of 700MB, and I noticed that over the 2min cycle only about 6 files got promoted(which is accepted), but the next cycle the remaining files start getting promoted, however we should not be allowing this as query should be reset after each cycle(attached are the logs, refer summeraised log to make life easy)


Version-Release number of selected component (if applicable):
================
3.7.5-9

Comment 2 Nag Pavan Chilakam 2015-12-14 08:36:19 UTC
Following was the detailed observation:
1)Checked that no files were heated for some time,and all were in cold tier with hot tier being empty. 
2)Now appened one line to 10 files ie bb{1..10}.mp so as to make them promote. Still the binary files in /var/run/gluster/goblin-tier-dht/promotequeryfile-goblin-tier-dht and /var/run/gluster/goblin-tier-dht/demotequeryfile-goblin-tier-dht were empty
3)Now the new cycle started (2min cycle)
4)the promote db file was changed to 4k size with demote being zero size.
File promotions started
5)After two minutes, ie when a new cycle started, only about 8 files had been promoted, with the promote binary still not reset.
=====> BUt what was observed was that on the other node(yarrow), the promote binary was reset and the demote binary was now 4k size <=====
Promotes and demotes were still happening in new cycle.

Attached are the logs

Comment 3 Nag Pavan Chilakam 2015-12-14 08:37:07 UTC
Created attachment 1105515 [details]
node#1 brick logs zod

Comment 4 Nag Pavan Chilakam 2015-12-14 08:37:57 UTC
Created attachment 1105516 [details]
node#2 brick logs yarrow

Comment 5 Nag Pavan Chilakam 2015-12-14 08:38:24 UTC
Created attachment 1105517 [details]
tier cli logs

Comment 8 Nag Pavan Chilakam 2015-12-22 07:11:44 UTC
We need to reset the binary file after every cycle

Comment 12 Shyamsundar 2018-02-06 17:42:55 UTC
Thank you for your bug report.

We are no longer working on any improvements for Tier. This bug will be set to CLOSED WONTFIX to reflect this. Please reopen if the rfe is deemed critical.


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