Scalability problems with EC2 driver
|Assignee:||Ruben S. Montero||% Done:|
|Category:||Drivers - Auth|
|Target version:||Release 1.4|
When running "a few" (problem encountered for 10 to 15) VMs, the EC2 drivers hangs (uses all CPU/memory) during polling and fails to
produce updated information. As the EC2 API tools are Java-based, one Java VM is started for each VM ran by ONE.
Possible solutions to this could be to tweak hardware allocation to Java VMs, or to poll all VMs in one call (ec2-describe-instances allows this).
fix tm/qcow2/ln fix clean before attach (#124)
Following this forum thread
Revert "fix tm/qcow2/ln fix clean before attach (#124)"
This reverts commit e99352f5d94043198087a8be6a48421221e5cf5c.
#1 Updated by Ruben S. Montero over 9 years ago
- Category set to Drivers - Auth
- Assignee set to Tino Vázquez
- Target version set to Release 1.4
I've limit the number of simultaneous EC2 operations to 5, so there are no than 5 java instances at the same time. This has been committed to the development branch. Needs to be tested
#2 Updated by riv nil over 9 years ago
1. I reverted back the changes committed by Ruben
2. I tried adding: EC2_JVM_ARGS="-Xms16m -Xmx64m" in [/etc/vmm_ec2/vmm_ec2rc]
No problem beyond 15 concurrent instances..
Evidently this is not an issue related to ONE rather EC2 setups..
Ruben, kindly revert-back the changes..
#4 Updated by riv nil over 9 years ago
Normally ec2 users are limited to a maximum of 20 concurrent instances.
Pls refer: http://developer.amazonwebservices.com/connect/entry.jspa?externalID=2157
This is fyi only..