Re: [users@httpd] balancer manager issue

[Suvendu Sekhar Mondal]

Felix,

On Wed, Aug 1, 2018 at 5:21 AM Felix Wong  wrote:
>
> using the built-in balancer-manager ui or curl to disable a balancer member 
> does not actually do anything.
>
> command: /usr/bin/curl --silent --insecure -o /dev/null -XPOST 
> 'https://localhost:443/balancer-manager?' -d b=home.monolith.on1.saasure.net 
> -d w=https://on1-lbmo01c.aue1t.internal -d 
> nonce=20e16be2-eb5f-42c0-b061-57af2968c7db -d w_status_D=1
>
> after running the above command, the status of the member reflects as 
> changed, but traffic is still flowing to the host (as evidenced by tailing 
> log files on the balancer member).
>
>  LoadBalancer Status for  balancer://home.monolith.on1.saasure.net  
> [pb9c23464_home_monolith_on1_saasure_net]
>   https://on1-lbmo01b.aue1t.internal   1.00  0  Init Ok   0  0  00
>  0
>   https://on1-lbmo01c.aue1t.internal   1.00  0  Init Dis   0  0  00   
>   0
>
Are you sure traffic is reaching to the disabled balancer member? As
per doc[1], Apache might automatically retry the member, but it should
not send anything. Are you seeing Elected count is changing? or both
Elected & Load count is changing in balancer manager UI for a member
after that change?

[1]https://httpd.apache.org/docs/2.4/howto/reverse_proxy.html

> balancer config:
> 
> # use our FQDN convention for BalancerMember URLs.
> # use the saasure.net domain because it maps to the public IP of an 
> instance.
> BalancerMember https://on1-lbmo01b.aue1t.internal timeout=320 retry=30 
> connectiontimeout=3000ms max=2000
> BalancerMember https://on1-lbmo01c.aue1t.internal timeout=320 retry=30 
> connectiontimeout=3000ms max=2000
> ProxySet maxattempts=2 lbmethod=byrequests
> 
> 
> # Explicitly turn off rewrites if we match /balancer-manager
> RewriteEngine off
> SetHandler balancer-manager
> Order deny,allow
> Deny from all
> Allow from 127.0.0.1 ::1
> Allow from localhost
> 
>
> has anyone seen this before, or is it just some subtle misconfiguration on my 
> side that i overlooked?
>
> thanks,
> felix
>

-
To unsubscribe, e-mail: users-unsubscr...@httpd.apache.org
For additional commands, e-mail: users-h...@httpd.apache.org

[users@httpd] Apache HTTPD not responding after running for several days

[Eric .]

Hi all,

I'm using Apache HTTPD 2.4.29 and 2.4.33 on Solaris 10 and 11. These HTTPDs
were compiled from source with the latest APR, APR-util and PCRE.
Upon starting, the HTTPDs initially run fine. After running for several
days to 2 weeks, HTTPD would stop responding. It wouldn't accept any new
connections although the processes are running. In the whole period the
incoming traffic volumn is low.
When in the "not responding" situation, stopping and then starting the
HTTPD would workaround the issue temporarily.


The symptoms are:
1. About 5 HTTPD processes are still running (checked by the "ps -ef"
command)
2. The TCP port is still being listened (checked by the "netstat -an | grep
LISTEN" command)
3. Any new incoming request would result in a connection in SYN_SENT state
(checked by "netstat -an") which usually disappears 1 minute later.


I have several Redhat 6.7 and Solaris 10 and 11 machines, and I find that
this problem only occurs on Solaris 10 and 11.

I'd appreciate your help. Thanks in advance.

Regards,
Eric