[jira] [Comment Edited] (HBASE-18946) Stochastic load balancer assigns replica regions to the same RS
[ https://issues.apache.org/jira/browse/HBASE-18946?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16278916#comment-16278916 ] stack edited comment on HBASE-18946 at 12/5/17 5:32 PM: Thanks [~anoop.hbase] for the intervention. I was thinking this progress and though I understood what is happening here but plain now that I did not. So back to square one. Can we give the replicas as a group to the LB and have it queue them as a group? I've not looked. But replicas being split over assignment queues w/o a global coordination -- and we do NOT want global coordination because we want the assignment queues to run w/o need of coordination so they run fast -- seems to be root problem? Will I have a go at it [~ram_krish]? <= Just because I've been in this AMv2 mess a good while now Thanks boss. was (Author: stack): Thanks [~anoop.hbase] for the intervention. I was thinking this progress and though I understood what is happening here but plain now that I did not. So back to square one. Can we give the replicas as a group to the LB and have it queue them as a group? I've not looked. But replicas being split over assignment queues w/o a global coordination -- and we do NOT want global coordination because we want the assignment queues to run w/o need of coordination so they run fast -- seems to be root problem? Will I have a go at it [~ram_krish]? Thanks boss. > Stochastic load balancer assigns replica regions to the same RS > --- > > Key: HBASE-18946 > URL: https://issues.apache.org/jira/browse/HBASE-18946 > Project: HBase > Issue Type: Bug >Affects Versions: 2.0.0-alpha-3 >Reporter: ramkrishna.s.vasudevan >Assignee: ramkrishna.s.vasudevan > Fix For: 2.0.0-beta-1 > > Attachments: HBASE-18946.patch, HBASE-18946.patch, > HBASE-18946_2.patch, HBASE-18946_2.patch, HBASE-18946_simple_7.patch, > HBASE-18946_simple_8.patch, TestRegionReplicasWithRestartScenarios.java > > > Trying out region replica and its assignment I can see that some times the > default LB Stocahstic load balancer assigns replica regions to the same RS. > This happens when we have 3 RS checked in and we have a table with 3 > replicas. When a RS goes down then the replicas being assigned to same RS is > acceptable but the case when we have enough RS to assign this behaviour is > undesirable and does not solve the purpose of replicas. > [~huaxiang] and [~enis]. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Comment Edited] (HBASE-18946) Stochastic load balancer assigns replica regions to the same RS
[ https://issues.apache.org/jira/browse/HBASE-18946?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16212288#comment-16212288 ] ramkrishna.s.vasudevan edited comment on HBASE-18946 at 10/20/17 7:38 AM: -- bq. As with some of regions with known location, there are new ones which needs to be assigned, these new ones could be assigned to the same region server which hosts the primary or other replica region. Yes agree with you. So i need to fix the issue that comes with this patch when used in EnableTableHandler first. bq.The previous logic is that when the first region is queued, it starts to wait assignDispatchWaitMillis to start the real work. With the patch, the whole batch is added at once, it skipped the addFirstOne logic. Will read and understand your comment and will be back here. was (Author: ram_krish): bq. As with some of regions with known location, there are new ones which needs to be assigned, these new ones could be assigned to the same region server which hosts the primary or other replica region. Yes agree with you. So i need to fix the issue with EnableTableHandler first. bq.The previous logic is that when the first region is queued, it starts to wait assignDispatchWaitMillis to start the real work. With the patch, the whole batch is added at once, it skipped the addFirstOne logic. Will read and understand your comment and will be back here. > Stochastic load balancer assigns replica regions to the same RS > --- > > Key: HBASE-18946 > URL: https://issues.apache.org/jira/browse/HBASE-18946 > Project: HBase > Issue Type: Bug >Affects Versions: 2.0.0-alpha-3 >Reporter: ramkrishna.s.vasudevan >Assignee: ramkrishna.s.vasudevan > Fix For: 2.0.0-beta-1 > > Attachments: HBASE-18946.patch, HBASE-18946.patch, > TestRegionReplicasWithRestartScenarios.java > > > Trying out region replica and its assignment I can see that some times the > default LB Stocahstic load balancer assigns replica regions to the same RS. > This happens when we have 3 RS checked in and we have a table with 3 > replicas. When a RS goes down then the replicas being assigned to same RS is > acceptable but the case when we have enough RS to assign this behaviour is > undesirable and does not solve the purpose of replicas. > [~huaxiang] and [~enis]. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Comment Edited] (HBASE-18946) Stochastic load balancer assigns replica regions to the same RS
[ https://issues.apache.org/jira/browse/HBASE-18946?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16194151#comment-16194151 ] ramkrishna.s.vasudevan edited comment on HBASE-18946 at 10/6/17 5:19 AM: - [~huaxiang] Thanks for the update. Great that you already feel this change will solve the problem. I was checking this issue but got pulled to something else. Will be back here and to check your suggestion. And yes we need to check all related areas and also other related branches. was (Author: ram_krish): [~huaxiang] Thanks for the update. Great that you already feel this change will solve the problem. I was checking this issue but got pulled to something else. Will be back here and to check your suggestion. And yes we need to check all related areas and also other related branches also. > Stochastic load balancer assigns replica regions to the same RS > --- > > Key: HBASE-18946 > URL: https://issues.apache.org/jira/browse/HBASE-18946 > Project: HBase > Issue Type: Bug >Affects Versions: 2.0.0-alpha-3 >Reporter: ramkrishna.s.vasudevan >Assignee: ramkrishna.s.vasudevan > Fix For: 2.0.0-beta-1 > > Attachments: TestRegionReplicasWithRestartScenarios.java > > > Trying out region replica and its assignment I can see that some times the > default LB Stocahstic load balancer assigns replica regions to the same RS. > This happens when we have 3 RS checked in and we have a table with 3 > replicas. When a RS goes down then the replicas being assigned to same RS is > acceptable but the case when we have enough RS to assign this behaviour is > undesirable and does not solve the purpose of replicas. > [~huaxiang] and [~enis]. -- This message was sent by Atlassian JIRA (v6.4.14#64029)