[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gluster-devel] Excessive memory usage with 1.3.12

From: Dan Parsons
Subject: Re: [Gluster-devel] Excessive memory usage with 1.3.12
Date: Wed, 5 Nov 2008 12:05:01 -0800

I haven't been monitoring gluster/zresearch very closely but I don't think their "commercialness" / community support effort has changed much since I started working with the software about 8 months ago. I think the reason io-cache has been getting a bit less love lately is because (just a guess) the devs are busy working on more often used features (io-cache isn't used very much, judging from the frequency it's mentioned on this list) and there are only so many devs to go around.

I've had a few other issues (which were mostly problems with my system architecture and not actual glusterfs code) and the devs (hey bulde!) have gone to heroic lengths to fix my stuff, spending hours with me on IRC and even once writing a custom patch to make gluster work in the non-standard way I was using it. You can't pay for support like that for any other product that I know of. As a suggestion, I've had faster/ more detailed support from their IRC channel (#gluster on freenode I believe). Also keep in mind that most/all of the devs are in India and are usually sleeping when us Americans are working. Though, that said, I've gotten help from them during our business hours.

I wish I had some help to offer regarding making it work on Solaris. Maybe the same thing is going on here as with io-cache, that is, the dev team is focusing on the most important core feature set to get 1.4.x ready. There is a company behind glusterfs (like I said in a previous email) but that doesn't mean they have unlimited developers. I'm guessing that Linux is their primary target OS and they are spending most resources getting everything working perfectly there, and will be able to spend more time on other systems once things stabilize for Linux (from what I know this is soon/"now").

This doesn't immediately help you, I know, but I suggest maybe giving glusterfs a bit more time/effort to do what you want.

Dan Parsons

On Nov 5, 2008, at 11:50 AM, rhubbell wrote:

Thanks Dan for the synopsis.

I wish I could experience a bug. (^;
I'm unable to even build the software on Solaris 10 Sparc.

I've seen a lot of software in production environments but when
a product has issues just compiling we usually don't put it on
our short-list of potential solutions.

I had some hope when I first found gluster and saw there was
activity.  But it seems with the commercial leaning of the product
that priorities are shifting. I've seen it in lots of other
projects. Just an observation, not making a morality or ethic

On Wed, 2008-11-05 at 11:43 -0800, Dan Parsons wrote:
I'm not really the best person to answer this question but I'll try

There is a commercial entity behind glusterfs - ZResearch(.com) of
which most/all of the developers are employed by. The developers are
(imo) very good and quick to respond to nearly every problem, it's
just this one particular issue where a response/fix has been a bit slow.

glusterfs is moving from 1.3.x to 1.4.x with some fundamental changes
involved but I don't think it's the same as what you mean by
"transitional state".

The product has been extremely stable for me (8gbit/s IO spread across
4 servers to 33 cpu nodes, bioinformatics work) and this memory "bug"
hasn't caused a crash under real work yet, just testing, but only
because our job input size is currently small.

So in summary, I love glusterfs, the devs/company behind it are solid,
it performs better for my work than others (pvfs, lustre) - it's just
this one io-cache memory bug that's been getting less-than-average
attention, and perhaps with this recent spark in attention that will
change :)

Dan Parsons

On Nov 5, 2008, at 11:20 AM, rhubbell wrote:

On Wed, 2008-11-05 at 19:55 +0100, Lukas Hejtmanek wrote:
On Wed, Nov 05, 2008 at 10:08:24AM -0800, Dan Parsons wrote:
Lukas, just to confirm your findings, I have the exact same
problem and
reported it about 2 months ago. Just like you, when all my stuff was
running under 32-bit, it wasn't an issue because of the 2GB limit,
now that I'm using 64-bit for everything, it is a potential system
crashing bug.

Yes, it's the same, unfortunately, I have no response from the
authors. So
nobody cares?

Well somebody cares.  Us.  But I am new here and wondering how
development on this project is funded. Is it all volunteer?
Partially funded via commercial offerings?

Is the project in a transitional state? Soon to go commercial?

Gluster-devel mailing list

Gluster-devel mailing list

Gluster-devel mailing list

reply via email to

[Prev in Thread] Current Thread [Next in Thread]