High latency for PUTs with read-while-writer enabled in clustered env

["Walsh, Peter" <Pe...@disney.com>]

Hello all,
We have a v3.0.1 full clustered environment for which we have configured read-while-writer according to Leif's blog (http://www.ogre.com/node/386), however we've noticed that PUT's are resulting in 4-7 seconds latencies while GET's behave as expected.  We've tried various PUT's for different origin servers to rule out origin server issue's as the cause.  Once we disable read-while-writer, the problem goes away.

Has anyone else experienced this issue with read-while-writer?  I didn't see any open bugs for this though I could have missed it.  We have disabled it in our production environment but ultimately would love to have this feature.

Thanks in advance,
Pete

RE: High latency for PUTs with read-while-writer enabled in clustered env

["Walsh, Peter" <Pe...@disney.com>]

Ok, we're continuing our testing but I'll open one with whatever data we collect.

Thanks,
Pete 


-----Original Message-----
From: Leif Hedstrom [mailto:zwoop@apache.org] 
Sent: Thursday, June 21, 2012 8:49 AM
To: users@trafficserver.apache.org
Cc: Walsh, Peter
Subject: Re: High latency for PUTs with read-while-writer enabled in clustered env

On 6/21/12 9:45 AM, Walsh, Peter wrote:
> Hello all,
> We have a v3.0.1 full clustered environment for which we have 
> configured read-while-writer according to Leif's blog 
> (http://www.ogre.com/node/386), however we've noticed that PUT's are 
> resulting in 4-7 seconds latencies while GET's behave as expected. 
>  We've tried various PUT's for different origin servers to rule out 
> origin server issue's as the cause.  Once we disable 
> read-while-writer, the problem goes away.
>
> Has anyone else experienced this issue with read-while-writer?  I 
> didn't see any open bugs for this though I could have missed it.  We 
> have disabled it in our production environment but ultimately would 
> love to have this feature.


Haven't heard / seen this one before. Please file a Jira ticket for it 
(and then attach your patches ;-).

-- Leif

Re: High latency for PUTs with read-while-writer enabled in clustered env

[Leif Hedstrom <zw...@apache.org>]

On 6/21/12 9:45 AM, Walsh, Peter wrote:
> Hello all,
> We have a v3.0.1 full clustered environment for which we have 
> configured read-while-writer according to Leif's blog 
> (http://www.ogre.com/node/386), however we've noticed that PUT's are 
> resulting in 4-7 seconds latencies while GET's behave as expected. 
>  We've tried various PUT's for different origin servers to rule out 
> origin server issue's as the cause.  Once we disable 
> read-while-writer, the problem goes away.
>
> Has anyone else experienced this issue with read-while-writer?  I 
> didn't see any open bugs for this though I could have missed it.  We 
> have disabled it in our production environment but ultimately would 
> love to have this feature.


Haven't heard / seen this one before. Please file a Jira ticket for it 
(and then attach your patches ;-).

-- Leif