incubator-cloudstack-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "edison su" <edison...@citrix.com>
Subject Re: Review Request: HA takes a lot of time to trigger HA if there are multiple storage pools in a cluster.
Date Wed, 25 Jul 2012 17:17:32 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/6133/#review9462
-----------------------------------------------------------

Ship it!


Ship It!

- edison su


On July 25, 2012, 9:09 a.m., Devdeep Singh wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/6133/
> -----------------------------------------------------------
> 
> (Updated July 25, 2012, 9:09 a.m.)
> 
> 
> Review request for cloudstack and edison su.
> 
> 
> Description
> -------
> 
> Bug: HA takes a lot of time to migrate VMs (trigger HA) to another KVM host if there
are multiple storage pools in a cluster.
> 
> The issue is as follows:
> 1. When CloudStack detects that a host is not responding to ping requests it'll send
a fence command for this host to another host in the cluster.
> 2. The agent takes a long time to respond to this check if the storage is fenced. This
is because the agent checks if the first host is writing to its heartbeat file on all pools
in the cluster. It is doing this in a sequential manner on all storage pool.
> 
> Making a fix to get rid of sleep, wait during HA. The behavior is now similar to Xenserver.
> 
> 
> This addresses bug 15646.
> 
> 
> Diffs
> -----
> 
>   agent/src/com/cloud/agent/resource/computing/KVMHAChecker.java b2df8c1 
>   scripts/vm/hypervisor/kvm/kvmheartbeat.sh c634655 
> 
> Diff: https://reviews.apache.org/r/6133/diff/
> 
> 
> Testing
> -------
> 
> Tested the fix by triggering HA on cluster with 4 storage pools. Before the fix an instance
was moved to another host int he cluster in more than 20 mins. With the fix the time is now
around 4 to 5 mins.
> 
> 
> Thanks,
> 
> Devdeep Singh
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message