[PATCH] Move elections to app3 and app4.

[Ricky Zhou]

Several people have seen gateway errors while voting.  Looking at the
proxy logs, I saw entries like:

(70007)The timeout specified has expired: proxy: error reading status line from remote
server app6.vpn.fedoraproject.org, referer:
https://admin.fedoraproject.org/voting/vote/18

and

(70014)End of file found: proxy: error reading status line from remote server
app6.vpn.fedoraproject.org, referer: https://admin.fedoraproject.org/voting/vote/22

So I'm guessing timeouts are being hit and connections are being closed
for some reason.  Since elections is being hit for several elections
now, I'd like to move it to app3 and app4 to avoid these issues.

The commit also removes the elections-server include from
app4.fedora.phx.redhat.com.pp since it's already included in appRhel.pp.
---
 configs/web/balancer.conf                     |    3 ++-
 manifests/nodes/app4.fedora.phx.redhat.com.pp |    1 -
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/configs/web/balancer.conf b/configs/web/balancer.conf
index ab4577b..0c27e63 100644
--- a/configs/web/balancer.conf
+++ b/configs/web/balancer.conf
@@ -96,9 +96,10 @@ ProxyRequests Off
 
 # admin.fedoraproject.org/voting/
 <Proxy balancer://electionsCluster>
+    BalancerMember http://app3.fedora.phx.redhat.com:80 loadfactor=1 timeout=10 retry=5
     BalancerMember http://app4.fedora.phx.redhat.com:80 loadfactor=1 timeout=10 retry=5
     #BalancerMember http://app5.vpn.fedoraproject.org:80 loadfactor=1 timeout=10 retry=5
-    BalancerMember http://app6.vpn.fedoraproject.org:80 loadfactor=1 timeout=10 retry=5
+    #BalancerMember http://app6.vpn.fedoraproject.org:80 loadfactor=1 timeout=10 retry=5
 </Proxy>
 
 # mirrors.fedoraproject.org
diff --git a/manifests/nodes/app4.fedora.phx.redhat.com.pp
b/manifests/nodes/app4.fedora.phx.redhat.com.pp
index 679286e..b66f1a2 100644
--- a/manifests/nodes/app4.fedora.phx.redhat.com.pp
+++ b/manifests/nodes/app4.fedora.phx.redhat.com.pp
@@ -3,5 +3,4 @@ node app4{
     include appRhel
     include vpn
     include supervisorCheckAppsEven
-    include elections-server
 }
-- 
1.5.5.6

[Nigel Jones]

Grrr,  I've been meaning to do this for a little while, lets just do it. +1
----- "Ricky Zhou" <ricky(a)fedoraproject.org&gt; wrote:

>  Several people have seen gateway errors while voting.  Looking at
> the
>  proxy logs, I saw entries like:
>  
>  (70007)The timeout specified has expired: proxy: error reading status
>  line from remote server app6.vpn.fedoraproject.org, referer:
>  https://admin.fedoraproject.org/voting/vote/18
>  
>  and
>  
>  (70014)End of file found: proxy: error reading status line from remote
>  server app6.vpn.fedoraproject.org, referer:
>  https://admin.fedoraproject.org/voting/vote/22
>  
>  So I'm guessing timeouts are being hit and connections are being
>  closed
>  for some reason.  Since elections is being hit for several elections
>  now, I'd like to move it to app3 and app4 to avoid these issues.
>  
>  The commit also removes the elections-server include from
>  app4.fedora.phx.redhat.com.pp since it's already included in
>  appRhel.pp.
>  ---
>   configs/web/balancer.conf                     |    3 ++-
>   manifests/nodes/app4.fedora.phx.redhat.com.pp |    1 -
>   2 files changed, 2 insertions(+), 2 deletions(-)
>  
>  diff --git a/configs/web/balancer.conf b/configs/web/balancer.conf
>  index ab4577b..0c27e63 100644
>  --- a/configs/web/balancer.conf
>  +++ b/configs/web/balancer.conf
>  @@ -96,9 +96,10 @@ ProxyRequests Off
>   
>   # admin.fedoraproject.org/voting/
>   <Proxy balancer://electionsCluster>
>  +    BalancerMember http://app3.fedora.phx.redhat.com:80 loadfactor=1
>  timeout=10 retry=5
>       BalancerMember http://app4.fedora.phx.redhat.com:80 loadfactor=1
>  timeout=10 retry=5
>       #BalancerMember http://app5.vpn.fedoraproject.org:80 loadfactor=1
>  timeout=10 retry=5
>  -    BalancerMember http://app6.vpn.fedoraproject.org:80 loadfactor=1
>  timeout=10 retry=5
>  +    #BalancerMember http://app6.vpn.fedoraproject.org:80 loadfactor=1
>  timeout=10 retry=5
>   </Proxy>
>   
>   # mirrors.fedoraproject.org
>  diff --git a/manifests/nodes/app4.fedora.phx.redhat.com.pp
>  b/manifests/nodes/app4.fedora.phx.redhat.com.pp
>  index 679286e..b66f1a2 100644
>  --- a/manifests/nodes/app4.fedora.phx.redhat.com.pp
>  +++ b/manifests/nodes/app4.fedora.phx.redhat.com.pp
>  @@ -3,5 +3,4 @@ node app4{
>       include appRhel
>       include vpn
>       include supervisorCheckAppsEven
>  -    include elections-server
>   }
>  -- 
>  1.5.5.6
>  
>  
>  _______________________________________________
>  Fedora-infrastructure-list mailing list
>  Fedora-infrastructure-list(a)redhat.com
>  https://www.redhat.com/mailman/listinfo/fedora-infrastructure-list

[Ricky Zhou]

On 2009-06-07 08:15:54 AM, Ricky Zhou wrote:
>  So I'm guessing timeouts are being hit and connections are being
> closed
>  for some reason.  Since elections is being hit for several elections
>  now, I'd like to move it to app3 and app4 to avoid these issues. 
Per some
discussion in #fedora-admin, we're considering this an outage,
so this change has been pushed.

Thanks,
Ricky