On 07/07/2014 03:23 AM, Cowe, Malcolm J wrote:
> 5) Is there any load-balancing capability with the HSM
framework?
The Coordinator will load balance requests across the copytool instances.
Hi Malcolm,
Thanks for your responses! However, I cannot see load-balancing in
affect at all. Does load balancing need to be explicitly turned on? If
so, how?
On my MDS, I've registered 2 agents, with same archive_id=5, each
running a single instance of posix copytool:
[root@hsm-mds1 lustre]# cat /proc/fs/lustre/mdt/lustrefs-MDT0000/hsm/agents
uuid=61f6bf53-2e22-d37b-45cd-fea234025701 archive_id=5
requests=[current:0 ok:11 errors:0]
uuid=516a502f-60d7-fa4e-09cf-ffe1ec43db3a archive_id=5
requests=[current:0 ok:0 errors:0]
Whenever I make a request to archive, the request goes to the *first*
registered one. As you can see from above, all 11 requests went to the
first one - thus no load balancing between the 2 agents.
If I kill the first copytool instance on agent 1, then the subsequent
archive requests will go to agent 2. However, that is not
load-balancing, it's more like failure recovery.
Please let me know how to correctly enable load-balancing, and/or what
is the load-balancing functionality available in 2.5.1.
Thanks,
-aamir