[f-nsp] ServerIron: direct requests to alternate server if all real servers are down.

[Drew Weaver]

Hi again list,

This has been working fairly well for awhile but I noticed that the connection counter for the server designated as backup seems to be incrementing:

State   CurrConn    TotConn

Now:

6          0   18836538

Yesterday afternoon:

6          0   18836021

Yesterday morning:

6          1   18835960

So you can see that the backup server has gotten about 578 connections in the past 24 hours but there haven't been any messages logged about the primary/active server being down due to healthcheck, reassign, or anything else.

Is there some kind of other metric the load balancer uses to decide when to send traffic to the designated backup? or will it send traffic to the backup if it sees a problem without logging any message about the primary server?

Thanks for any advice on this matter.

-Drew





From: Scott T. Cameron [mailto:routehero at gmail.com]
Sent: Tuesday, April 19, 2011 1:42 PM
To: Drew Weaver
Cc: foundry-nsp at puck.nether.net
Subject: Re: [f-nsp] ServerIron: direct requests to alternate server if all real servers are down.

There's a lot of variables involved with that question.  Depends on how the server is going down (loss of power, health check, etc).  Because there are so many different reasons it could take a large amount of time to come back online, any suggestions are bad guesses.

Things to keep in mind:

- you may have no-fast-bringup globally or on a specific port
- your health check may be waiting on something else to bring up the port
- you may have sticky persistence

etc.

Scott
On Tue, Apr 19, 2011 at 8:01 AM, Drew Weaver <drew.weaver at thenap.com<mailto:drew.weaver at thenap.com>> wrote:
Hi Scott,

Thank you for that tip that got me where I needed to be.

My last question is it seems to take rather a long time (3-4 minutes) for it to switch back to the main servers after they come back up.

Is there some way to make that faster?

thanks,
-Drew


From: Scott T. Cameron [mailto:routehero at gmail.com<mailto:routehero at gmail.com>]
Sent: Monday, April 18, 2011 8:38 PM
To: Drew Weaver
Cc: foundry-nsp at puck.nether.net<mailto:foundry-nsp at puck.nether.net>
Subject: Re: [f-nsp] ServerIron: direct requests to alternate server if all real servers are down.

backup is correct.

You need to add 'port # lb-pri-servers' on the vserver.

Scott
On Mon, Apr 18, 2011 at 8:19 PM, Drew Weaver <drew.weaver at thenap.com<mailto:drew.weaver at thenap.com>> wrote:
Hi,

How can you make a serveriron XL send requests to a different server only if the primary server(s) fail healthchecks?

I tried setting the secondary server as a backup server but the load balancer was balancing traffic between the primary servers and the 'Backup server' I only want requests to be sent to the backup server in the event the other real servers are offline.
Thanks,
-Drew

_______________________________________________
foundry-nsp mailing list
foundry-nsp at puck.nether.net<mailto:foundry-nsp at puck.nether.net>
http://puck.nether.net/mailman/listinfo/foundry-nsp


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/foundry-nsp/attachments/20110611/857715c6/attachment.html>