GNOME.org
 

Re: [Evolution] Evolution 1.4.0 *very* slow

There are upwards of 9000 emails on the server, more than a third of
which are from various Redhat, Ximian and other Linux-related mailing
lists.  I was asked to send a few hundred lines of the log file; when
the log file reached a little over 400 lines - I canceled. I wasn't
aware that I should have sent more than I did.  In fact, I don't know
enough about anything Linux-related to be able to determine what you
might need beyond that which I'm asked to send - and I'm not afraid to
admit it.


There is an adage which goes: the proof is in the pudding.  My email is
working very well using Evo 1.0.8-11 on Redhat 8.0.  I tried Evo 1.2.4
(?), and the various 1.4 versions - all exhibited the same behavior:
glacially slow retrieval times. When I tried 1.2.4 (?), I had roughly
5800 emails on the server; after first retrieval, subsequent attempts
could not be completed within 24 hours (completing once over the course
of a weekend).  1.4 was better, taking 'only' 7 hours to do subsequent
retrievals on a little more that 6400 emails.


There have been _no_ changes to our email server (I know, I run the data
center).  No service packs, no software upgrades, no hub or switch
changes, cabling is precisely the same.  Log files do not indicate any
abnormal activity (neither the mail server nor the firewall).  We have a
few in-house folks and roughly 100 off-site; none of them experienced
delays; that may or may not tell you anything, since I'm the only one
using Linux.


You know, this is really fascinating to me - being a professional
programmer (20 years this past May), I don't often get to see what the
results of my work are until I deploy my code.  I test as best I can,
using the same code, systems and datasets that my users will be using. 
Many times I only learn of a problem either by user reports, or by
checking some sort of log file.


You guys have it much harder in that you have no way of knowing exactly
what environment your product will be run in; you can make certain 
assumptions, but that's all - nothing's a given.  Your problem
resolution is basically the same as mine: test as best you can, listen
to user reports, check log files. But remember that you know your
product better than your users do (or you should!). If you need
something from them to help you in tracking down a problem, don't assume
that they know what you need without you specifically asking for it. 
Linux, and its derivative symbiotics, is attracting new (_really_ new)
users every day; many of them have no clue about how to do anything
beyond Windows or mainframes and will need plenty of hand-holding.
Smugness and defensive positions serve no one well (not to imply that
you're either smug or defensive), and only harm the product and the user
experience.


It's really strange experiencing a problem from a "dumb" users
perspective. All in all I consider this a valuable learning experience,
one that will hopefully be of benefit to my user community.


When I get over my flu and feel up to fighting the battle again, I may
install RH9 / Evo 1.4 again.  Perhaps I'll even be foresighted enough to
run a CAMEL_VERBOSE_DEBUG=1 log on the RH8/Evo 1.0.8-11 combo that I'm
using now.  But not today.


On Fri, 2003-06-27 at 00:03, Jeffrey Stedfast wrote:



the only thing in that log that you provided was:

<connect>
send: CAPA
recv: -ERR ...
send: USER <user name>
recv: +OK
send: PASS <passwd>
recv: +OK
send: LIST
recv: <long list of emails... and before it finished, you hit cancel>

I gather you hit cancel because it was taking such a long time and you
figured that whatever was in the log would be enough to show us what was
so slow...

however, what was slow was just the server being slow in sending us the
list of messages you had stored on your pop server (hundreds of them).

there's nothing we are doing in that transaction that suggests we are
doing anything inefficiently.

Jeff

On Thu, 2003-06-26 at 17:47, Tom wrote:

Well... I filed the report, and eventually got an email back from
someone about it.  I've forgotten the bugzilla bug number, and have
apparently deleted the emails that had it on there.

The gist of the bugzilla reply was that it is a network latency issue,
in effect the network is taking too long to do whatever it needs to do
in order for me to receive my email in a timely manner. No problem,
nothing would be done, issue resolved.

I've since removed RH9 and Evolution 1.4, and re-installed RH8 and
Evolution 1.0.8-10 (since upgraded to 1.0.8-11).

Guess what - everything is working like a charm.  Email is lickety-split
on both sending and receiving.

In the end, I don't think it's a network issue at all. I'm even more
convinced that it's programming issue, if only because this is the 3rd
time that I've done this retrograde with exactly the same result.

On Tue, 2003-06-24 at 21:45, David Johnston wrote:



On Mon, 2003-06-16 at 11:19, tomh simas com wrote:
Tom, the easy way to debug Evolution is this:
1. Click on the Red Hat menu
2. Click on System Tools; you get a new menu
3. Click on Terminal; after a pause, you get a window that looks like a
terminal.
4. Type "export CAMEL_VERBOSE_DEBUG=1" without the quotes and hit the
enter key
5. Type "evolution" without the quotes and hit the enter key.

Hope this helps!
-David

[RH9.0]

So I would need to "gedit .bash_profile" and add a line which reads

"export

CAMEL_VERBOSE_DEBUG=1"?





                      Jeffrey Stedfast
                      <fejj ximian com>            To:

tomh simas com

                      Sent by:                     cc:       Not Zed

<notzed ximian com>, evolution lists ximian com

                      evolution-admin lists        Subject:  Re:

[Evolution] Evolution 1.4.0 *very* slow

                      .ximian.com


                      06/16/2003 10:05 AM






if you are using bash (which you probably are - since I'm pretty sure
most distros use this by default), you can set an environment

variable

using 'export'. For example:

export CAMEL_VERBOSE_DEBUG=1

Jeff



How do I set CAMEL_VERBOSE_DEBUG = 1 in the environment?





_______________________________________________
evolution maillist  -  evolution lists ximian com
http://lists.ximian.com/mailman/listinfo/evolution

--
Jeffrey Stedfast
Evolution Hacker - Ximian, Inc.
fejj ximian com  - www.ximian.com
[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]