This module appears to break the "backup" upstream functionality

[adamdamico]

Hello,

I have patched the 0.8.54 SRPM from EPEL with this module and run some tests. When I specify a health check according to the documentation, and also specify one of my upstreams as a "backup," the backup never kicks in. At the point where both of the non-backup upstreams are down, any request results in error 502 and the backup is untouched.

I have tried both a check_http_send and a tcp health check. Commenting out the health check directives allows the backup upstream to function normally.

I was really hoping this would work. My goal here is to have deep health checking for my normal upstreams, and also have a degraded-mode light-weight upstream as a last resort.

Any clue why this might be occurring?

[yaoweibin]

Hi Adam,

That's my fault. It's a bug. I have fixed it in the latest revision. The
bug appears in the patch file. Please add the patch again.

Thanks very much for this report.

On 2011-6-2 8:41, adamdamico wrote:

Hello,

I have patched the 0.8.54 SRPM from EPEL with this module and run some tests. When I specify a health check according to the documentation, and also specify one of my upstreams as a "backup," the backup never kicks in. At the point where both of the non-backup upstreams are down, any request results in error 502 and the backup is untouched.

I have tried both a check_http_send and a tcp health check. Commenting out the health check directives allows the backup upstream to function normally.

I was really hoping this would work. My goal here is to have deep health checking for my normal upstreams, and also have a degraded-mode light-weight upstream as a last resort.

Any clue why this might be occurring?

Weibin Yao