Re: Memcached Response Time

Thu, 12 Sep 2013 04:24:09 -0700

Thanks for data Paolo.

The problem was in the response read - it tried to fill buffer which was larger then necessary. The read was then waiting till the timeout expired - although the response was in the buffer already.

The attached patch solves the problem (will be fixed in the next monit version too).

Regards,
Martin

Attachment: memcache.patch
Description: Binary data




On Sep 12, 2013, at 11:45 AM, Paolo Martinelli <[email protected]> wrote:

Hi all,

I assume the protocol is TCP anyhow we are using the monit implemetation of the memcache protocol.
I have no clue on the possible impact of the Timeout value, anyhow even if the response time is always 5 seconds we are not receiving any alert.

From a first look at the network trace it seems that the check on the memcached availability is done every 6 seconds even if in the configuration there is "set daemon 60".
All the other checks (MySQL, HTTP, etc.) are on a 60 seconds frequency.

Cheers,

Paolo

P.S.: I'm sending the network trace on a separate mail.

From: "sven falempin" <[email protected]>
To: "This is the general mailing list for monit" <[email protected]>
Sent: Wednesday, September 11, 2013 2:54:44 PM
Subject: Re: Memcached Response Time

Is there a relationship with the Timeout value ?


On Wed, Sep 11, 2013 at 5:28 AM, Martin Pala <[email protected]> wrote:
Hi,

please can you provide network trace of the communication between Monit and memcached? (you can use for example tcpdump or wireshark to catch the packets - let me know if you need help with it).

Do you use TCP or UDP for the memcached test? 


Regards,
Martin



On Sep 11, 2013, at 10:05 AM, Paolo Martinelli <[email protected]> wrote:

Hi all,

I have just deployed monit across different server in my farm.
We have multiple installation of memcached and all of them are kept monitored with a script like this:

check host memcache_host with address x.x.x.x
    if failed port y and protocol memcache then alert

The "strange" thing that I see is that the response time from any memcached instance is always 5 seconds. 

Do you know if there is any specific reason for this behavior?

Many thanks,

Paolo

--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general


--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general



-- 
---------------------------------------------------------------------------------------------------------------------
() ascii ribbon campaign - against html e-mail 
/\ 

--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general

--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general

--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general

Reply via email to