cloudstack-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Wilder Rodrigues (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CLOUDSTACK-4374) Enable HA for routers that are part or a redundant network or VPC
Date Thu, 10 Dec 2015 10:51:11 GMT

    [ https://issues.apache.org/jira/browse/CLOUDSTACK-4374?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15050573#comment-15050573
] 

Wilder Rodrigues commented on CLOUDSTACK-4374:
----------------------------------------------

I will fix this.

Redundant Routers is not the same as Haigh Available. What makes a network redundant is having
2 routers that are managed by a third part application, for instance keepalived. Having a
router HA is actually saying that the given router will be controlled by the  High Availability
monitor. Hence fix any problem we might face.

To sum it up: high availability and redundancy are 2 different things, conceptually.

Cheers,
Wilder

> Enable HA for routers that are part or a redundant network or VPC
> -----------------------------------------------------------------
>
>                 Key: CLOUDSTACK-4374
>                 URL: https://issues.apache.org/jira/browse/CLOUDSTACK-4374
>             Project: CloudStack
>          Issue Type: Bug
>      Security Level: Public(Anyone can view this level - this is the default.) 
>          Components: Virtual Router
>    Affects Versions: 4.1.0
>            Reporter: Roeland Kuipers
>            Assignee: Wilder Rodrigues
>
> We provide redundant routers with HA functionality through a special service offering.
> However these router pairs are provisioned with ha_enabled=0, so when one or both of
them fail they will never be restarted by CS. 
> 2013-08-16 15:51:16,101 DEBUG [cloud.ha.HighAvailabilityManagerImpl] (HA-Worker-0:work-4335)
VM is not HA enabled so we're done.
> This is currently hardcoded in VirtualNetworkApplianceManagerImpl.java @ 1633
>    boolean offerHA = routerOffering.getOfferHA();
>                 /* We don't provide HA to redundant router VMs, admin should own it all,
and redundant router themselves are HA */
>                 if (isRedundant) {
>                     offerHA = false;
>                 }
> We like redundancy and like to have HA on our redundant routers. We like to configure
this ourselves through service offerings and do not like being helt hostage by these lines
of codes:) We do like to own it all in our admin role :)
> Besides this, this is also very counter-intuitive as we were expecting HA on our redundant
routers, since it was configured on their service offering.
> So can we get rid of these lines of code? And have this controlled through service offerings
as it should IMHO.? Unless this has negative impact which we are not aware off?
> Cheers & Thanks,
> Roeland
> Details of the original commit which injected this code:
> Commit: a269b089ae38d0d04db2fa0f4c8e839480476ddc [a269b08]
> Parents: a2cc66ce41
> Author: Sheng Yang <sheng.yang@cloud.com>
> Date: 17 december 2011 03:52:59 CET
> Commit Date: 19 december 2011 22:29:48 CET
> bug 12608: NaaS: Don't shutdown elements if cleanup=false
> We can use the restartNetwork mechanism to recover the disconnected redundant
> router.
> Also disable HA for redundant router. Admin would take responsibilty to recover
> the failure router, because redundant routers themselves are one layer HA.
> status 12608: resolved fixed



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message