Topic: Slow reply/forward response

[David Waldock]

Hi

I recently installed a relatively high-end system for a client (512MB RAM, 2GHz Processor, 120MB HDD, 100Mb/s NIC).  I ran SME on a MUCH lower spec system, and was always impressed by it's response times.

However, my client has experienced (and I can confirm) that it can take *forever* (or at least a very long time) for the "reply/reply to all/forward" window to be ready to accept a message.

Does anyone have any suggestions?

David

[Dick Morrell]

Question: why if you installed a non supported product at a client site do you expect support. Wouldn't it do more for your reputation with the client to install MAS ? Then have support ?

Seems a bit lame to put a GPL non supported solution in a corporate environment. I wouldnt do it - would just lower my value add and make me look pants.

Dick

[Michiel]

> Question: why if you installed a non supported product at a
> client site do you expect support. Wouldn't it do more for
> your reputation with the client to install MAS ? Then have
> support ?

In theory yes, but it's more hassle to become a Mitel reseller than to customize a SME distro...

[Joe]

Dick,

Why do you keep on about these sorta things, this is a technical fourm not to complain about why this, should have this - o boo hoo.

You havent posted anything since answering somone's thread last week, even then it was to complain.

(I'm not gonna start another thread fight, so i'll end it now) anyway looks like Dick has he's hands full with another thread.

[Dick Morrell]

Oh I agree with that lol...

A Mitel reseller asked me to go on a course. I could take the bloody course not just sit on it. I did giggle and tell him to shove it.

Seriously - why sell SME when its unsupported and so insecure ? SSL Apache Perl and other bits remain unpatched. It's not really something I'd want my customers using - MAS is updated and it pays for SME. If Mitel cant get the mix right just use RedHat, hardly tricky.

At least if on RHN it gets updated...

Dick

[Dick Morrell]

Joe thats poo

I've posted 11 times since.

It's not complaining its just pointing out how comical this system is

and daft thing is you all say it too.

[SloopJohnB]

It's beginning to sound like SME and MAS are related, one is an arrogant bast... and the other an unwanted step-child. Look, a product, whether or not it makes profit, is a product. Be it a service, an item, or a warranty, it is still your product. It sounds to me that the MAS/SME reps here in this forum, do not wish to "waste their time" dealing with these "petty" questions from this forum, as this forum doesn't bring home the immediate cash every week. Instead of looking at the short-term benefits of SME to MAS, look at the cost savings you have in developing a GPL product with thousands of users for free! You don't have to contact your exisitng clients and ask if they want to be in a beta program, and you don't have to service them, all you have to do is respond to posts in this forum and develop strategies and new modules for future releases (plus any bug fixes). Think of how much you would have had to pay for this R&D just 10 years ago, or when there was no internet. Think of your SME clients (providing beta testing) as clients not freeloaders wasting your time.
If you feel I am wrong, please don't hesitate to rebut.
SloopJohnB

[Michael Doerner]

An idea for David:

Just a guess - it  might default to start the smp (multiprocessor) version but actually be a single processor system?

Regards,
Michael Doerner

[dave]

David,

You might try running a ps -ef and see if there's anything running that's taking all the processor or memory or both.  I know I've seen it here on the boards but I don't recall the specifics.  Take a look at your logs too, see if there's anything being logged that might point to a hardware issue.  I too have run SME on a very low end system and saw no real performance issues.

As for those that are here only to complain ... Well, everyone is entitled to their own opinion.  I have mine too.  BTW:  What are you expecting from free software?  That the Mitel guys respond at all to issues regarding unsupported software is exceptional.

[David Waldock]

The client in question is a small charity which needed a flexible email/file server solution.

I of course did not sell it to them (as that would be contrary to the GPL), I provided them with the most cost effective solution available to them for less than £500, which needs minimal administrative support (for which they will have to pay).

Anyway, none of this answers the questions: is this a known problem with certain hardware?

David

[Terry]

As Michael said, is the correct version of software being selected at boot up?  SMP vs Uni processor.

[Charlie Brady]

David Waldock wrote:

> I of course did not sell it to them (as that would be
> contrary to the GPL),...

No, it wouldn't.

You can learn more about the GPL by reading it (as you could have done, for instance, at installation time), and by reading FSF's Fequently Asked Questions:

http://www.gnu.org/licenses/gpl-faq.html

Charlie

[dave]

David,

You may have already checked this...

I have read here on the boards several times about certain types of IDE drives not properly configuring for SME.  I'm no expert but the info I recall seeing had to do with running hdparm or something like that to force the correct mode.  If the HDD is not running in the correct mode, access times can be really slow.

Dave

[David Waldock]

My understanding from the documentation is that where the SMP kernel is installed, it also installs the UP kernel and give the two options at the boot menu. In this instance, it only has one option at boot, which I can only assume to be UP.

I do have to say, however, that once it gets past the LILO-like bootscreen, the text whizzes by so fast that I wouldn't be able to tell you if SMP appears on the screen.  Is there another way to verify that SME isn't running SMP automagically?

Thanks for the idea though.

David

[David Waldock]

Apologies for the length of this email in advance.

> You might try running a ps -ef and see if there's anything
> running that's taking all the processor or memory or both.  I
> know I've seen it here on the boards but I don't recall the
> specifics.  Take a look at your logs too, see if there's
> anything being logged that might point to a hardware issue.
> I too have run SME on a very low end system and saw no real
> performance issues.

I now have off-site SSH access, making this somewhate easier!

A ps -ef reveals nothing which appears to be taking up anything more that 00:00:01 processor time, with the sole exception of (squid) -D running under uid squid, which is taking up a measly 00:00:07.

I also made a note of the load average: 0.02, 0.01, 0.00, which strikes me only as exceptional in that it is so low.

boot.log is boring in it's SUCCEEDED messages, although i am seeing the occassional ^[[60G message which I don't recognise.

At least one instance of Individual machine logs show:

[2003/10/13 21:35:30, 0] lib/util_sock.c:read_data(436)
  read_data: read failure for 4. Error = Connection reset by peer

There also several instances of the following error messages in the httpd log:

[Wed Oct 22 01:22:37 2003] [error] [client 192.168.8.4] File does not exist: /h
ome/e-smith/files/primary/html/_vti_inf.html
[Wed Oct 22 01:22:37 2003] [error] [client 192.168.8.4] File does not exist: /h
ome/e-smith/files/primary/html/_vti_bin/shtml.exe/_vti_rpc
[Wed Oct 22 03:59:17 2003] [notice] child pid 31678 exit signal Segmentation fa
ult (11)
[Wed Oct 22 04:15:11 2003] [notice] child pid 31680 exit signal Segmentation fa
ult (11)
[Wed Oct 22 05:15:39 2003] [notice] child pid 711 exit signal Segmentation fault (11)
[Thu Oct 23 04:56:52 2003] [error] mod_ssl: SSL handshake timed out (client 62.
255.177.95, server www.bucksgaylink.org.uk:443)
[Thu Oct 23 23:12:33 2003] [notice] caught SIGTERM, shutting down
[Thu Oct 23 23:13:37 2003] [notice] Apache configured -- resuming normal operat
ions
[Thu Oct 23 23:13:37 2003] [notice] suEXEC mechanism enabled (wrapper: /usr/sbi
n/suexec)
[Thu Oct 23 23:13:37 2003] [notice] Accept mutex: sysvsem (Default: sysvsem)
[Thu Oct 23 23:17:24 2003] [notice] caught SIGTERM, shutting down
[Thu Oct 23 23:18:29 2003] [notice] Apache configured -- resuming normal operat
ions
[Thu Oct 23 23:18:29 2003] [notice] suEXEC mechanism enabled (wrapper: /usr/sbi
n/suexec)
[Thu Oct 23 23:18:29 2003] [notice] Accept mutex: sysvsem (Default: sysvsem)
[Thu Oct 23 23:19:37 2003] [error] [client 192.168.8.6] File does not exist: /h
ome/e-smith/files/primary/html/mail


Any hints would be appreciated.

David