Re: UDP ascii protocol: quit command causes hang

Hi Ayrat: Thanks for reporting this. This is a known issue (see http://code.google.com/p/memcached/issues/detail?id=158) for which I have fix and one of these days I will commit said fix to the trunk. Thanks! Eric On 2/3/11 4:55 AM, Ayrat Khalimov wrote: > Hi, All, > > Here is the command "q

Re: Issue 158 in memcached: Single packet DoS on UDP channel

Yeah, the problem is due to how the server deals with a "large" (over 1024 bytes IIRC) malformed UDP messages (in this case malformed means no \r\n). I've had a fix for a while, just have not gotten around to getting a unit-test written and checking in the fix. On Dec 14, 7:20 am, memcac...@google

Re: REMOVE ME

On 12/7/10 9:59 PM, pravesh suyal wrote: rm -rf pravesh http://groups.google.com/support/bin/answer.py?hl=en&answer=46608 > > > ** > *pravesh suyal* > > > --- On *Tue, 12/7/10, Jakub Łopuszański //* > wrote: > > > From: Jakub Łopuszański > Subject: Problems with increment/decrement with

Re: Force memcached to write to all servers in Pool for reliability

Hi Jay: Trond Norbye, who added this feature into libmemcached, wrote a blog entry describing how to use libmecached for replication. You can find it at http://blogs.sun.com/trond/entry/replicate_your_keys_to_multiple Good luck Eric Jay wrote: Hi Brian, I don't know if you have received my

Re: Memcached set is too slow

PHP5.3, libmemcached 0.35, memcached 1.0.0 Is this really the version of the memcached server you are using (1.0.0) If so, that is certainly out-of-date. Latest version is 1.4.*. Eric Chaosty wrote: We have found that Memcahed::set stores items around 100-200kbs for 0.10-0.11 seconds, its t

Re: [Fwd: edging toward pandora-build]

". If you think such a change would be too pervasive/invasive and therefor a bit difficult to review, then I say "God Speed" to your approach :-) Eric Matt Ingenthron wrote: > > Eric Lambert wrote: >> Hey Matt, I don't have a problem with the strategy you've outli

Re: [Fwd: edging toward pandora-build]

Hey Matt, I don't have a problem with the strategy you've outlined below but just out of curiosity how pervasive would the changes be if you submitted them in larger chunks as opposed to these small type changes you suggest? Eric Matt Ingenthron wrote: > Hi all, > > I'd sent this out a few weeks

Re: Newbie Question

> > Memcache doesn't have any replication and doesn't need it. Its a > distributed cache, which means the keys will be distributed > automagically between machines. While the server does not perform any replication I believe the libmemcached client does provide replication functionality. Joseph

Re: to be added to the memcached tools page

added luciano11 wrote: Hey guys -- I saw the memcached advanced reporter release post fly by today. Probably should toss this up on the memcached tools page: Advanced Reporter - a stats tool for memcached which provides information about the most active clients and object keys for your memcac

[Fwd: Hi,Could you help me?A new memcached client for java]

Hi Dennis: Forwarding this to memecached mailing list. Eric --- Begin Message --- hi,eric I have written a new memcached client for java,called xmemcached (see http://code.google.com/p/xmemcached/) It's main features: 1、Supports all memcached text based p

Re: memcached tool links/descriptions

added ... luciano11 wrote: Please add to the tools page. Feel free to shorten the descriptions as necessary. brutis http://code.google.com/p/brutis/ Brutis is a tool designed to exercise memcache instances by providing reproducible performance data for comparison purposes. statspr

Re: For review: Stats slabs should include the number of requested bytes

I've modified trond's patch to also calculate the chunk space 'waste_rate', if anyone is interested http://gist.github.com/92656 Dustin wrote: On Apr 8, 8:54 pm, Eric Lambert wrote: Fair point so this raises a question about who is the audience for the sta

For review: Fix for issue 38 (server does not respond to binary protocol udp requests)

== issue == Clients who sent udp request using the binary protocol would never receive a response from the server. http://code.google.com/p/memcached/issues/detail?id=38 == patch summary == Prior to this patch, the connection structures that were tied to UDP sockets always set the connection

Re: For review: Stats slabs should include the number of requested bytes

convert it to a lossy calculated value... It's useful to be able to write a monitor to watch the more exact ratios, or graph exact sizes. Which you lose if you calculate down the information too far. That said ... I don't have any great alternative ideas offhand. -Dormando On Wed, 8 Apr

Re: For review: Stats slabs should include the number of requested bytes

Looks fine to me. Although perhaps we could use a better description. The text 'mem_requested' is a little unclear to me. May be we call it "bytes_used_in_slab" but that's a bit wordy Also I wonder if we should represent this value a little differently. So the point of this is to

Re: Memcached Hackathon at MySQL Users Conference: tentative date April 21st

FYI, due to some scheduling issues we've changed the date we have request for the Hackathon from Tuesday April 21st 7:30 PM to Wednesday, April 22nd, 8:30 PM. This is still tentative as the folks running the conference have not finalized the BOF schedules. Eric On Mar 31, 8:04 pm, Eric La

Memcached Hackathon at MySQL Users Conference: tentative date April 21st

Howdy: We are putting together a Memcached Hackathon at this year's MySQL Users Conference. The tentative date and time is Tuesday, April 21st at 7:30 PM. The location will be the Hyatt Regency Santa Clara, 5001 Great America Parkway, Santa Clara, CA 95054. Date and room number wont be finaliz

Re: Confused by "ceiling" vs "max" memory size for memcached service

Hi: I am not at all familiar with jmemcached, but I took a quick look at the code and from what I can tell, max memory defines the total number of bytes allocated to the cache and the 'ceiling' is used to control when items will be evicted from the cache. Once the total amount of data in th

patch: ascii set operations fail when using udp

The problem appears to be due to the fact that the complete_nread() function does not handle the udp protocol (c->protocol == ascii_udp_protocol). See proposed fix and test attached. FWIW, glancing at the code suggest that there is no support for any UDP operations using binary protocol. Will cr

Re: For review: issue 28

Hey Trond: I was a little unclear on the -R text. Here is my attempt at rewording. This option seeks to prevent client starvation by setting a limit to the number of sequential requests the server will process from an individual client connection. Once a connection has exceeded this value, th

Re: spy client under heavy load

Sure, I'll put it on my github account in the next couple days Moazam Raja wrote: Eric, would it possible for you to publish your test setup so others can try it on different hardware and/or tweak the config a bit? Thanks. -Moazam On Fri, Mar 6, 2009 at 8:22 AM, Eric Lambert

Re: spy client under heavy load

Hey Adam: I've done some simple profiling of the spy client and noticed a drop off in performance when the number of threads exceeds about 40 (this was on a Sun Fire x2200 m2, dual core) which sounds similar to the issue you are seeing. You can see the details of the benchmark on my blog ht