Here, with his permission, is the start of an email exchange with
Howard Butler at umn...
I certainly have been known to complain about the treadmill of
upgrades that was (once) Plone. Version 3, not that it helps you
right now, is MUCH more stable and I know from what I've seen and what
I heard at the recent conference that our cries of "enough already"
have been heard. I was able to upgrade some relatively simple 2.5
sites to 3, but the remaining 2.1 site (which runs a custom election
system) will probably remain for some time; I always imagined I would
let old sites run old software unless someone screamed.
Since you say you're a volunteer, I take it there isn't much money to
spend on this. Is there a group at UMN that provides virtual
servers? Any chance of moving your site to a faster machine? A few
groups have gotten Plone running on Amazon EC3, and there are now
established Plone/Zope hosting services you might be able to use, if
you had some money. You might then be able to offload some of the
server admin duties.
Have you tried the plone-users list or IRC to try to tackle the
external editor issues, or perhaps the PHC upgrade problems you ran
into? It sounds like there is a lot going on in your site that we
would only be able to pick away at the problem bit by bit.
I don't think you offended anyone... at least not me, maybe because
I've done a bit of venting on the plone-users list myself this past
week about performance issues.
On Nov 2, 2008, at 9:07 PM, Howard Butler wrote:
On Nov 2, 2008, at 6:33 PM, T. Kim Nguyen wrote:
Hi Howard - your page at http://mapserver.gis.umn.edu/development/rfc/ms-rfc-46
came up on one of the Plone mailing lists. I've been spearheading
the Plone rollout on our campus at Univ. of Wisconsin Oshkosh and
thought I'd ask if you thought there was a way I might be able to
help you with your Plone performance issues. I noticed you mention
having older (less reliable?) hardware: what exactly is running
your Plone site, and is Solaris a must? The Zope installation docs
mention that the threading model in Solaris will slow things down
relative to running a different OS on the same hardware, though I
found in my testing a couple of years ago it was maybe a 10-15%
difference only - not horrible. Also, you're running Plone 2.5,
which could be a factor since Plone 3 is 50% faster for some
things. Are you running CacheFu and a proxy like Squid?
Anyhow, let me know if you think I can help or if it's a lost
cause. You can check out what we're doing at UW Oshkosh here:
I have read through the thread on the evangelism list thread about
this and will group some responses to all of it here.
- We are running Plone 2.1 with an approximately Dec 2005
PloneHelpCenter, a custom gallery archetype product (written by me),
and a few other products (PDF, Questionnaire, etc)
- We have older CacheFu and a squid instance going, which allows us
to at least get past search engine crawler day without imploding :)
- Slowlaris obviously wasn't our first choice when looking to run
the site, but that is what we had available. A few years down the
road, there are now more options in that respect.
I attempted to get this stuff to 2.5 about a year and a half ago,
and after a good bit of time struggling, I gave up. I know that a
lot of the pain is self-inflicted. I used SVN versions of things
like PHC, etc, that makes migration especially painful. At the time
the website was rolled out, it did fill our project's needs fairly
well, and moving our docs from docbook to ReST has been probably the
most beneficial aspect of the entire effort. Our problems are
mostly related to the fact that there is only a single admin, our
site hasn't aged very well with respect to staying on the Plone
upgrade mill, and the rest of the developers/users never really
bought into the concept of the site. They were fine editing files
through cvs/svn and having a process generate some html for them.
When my RFC talks about the site not being successful in generating
through-the-web editing, the developer buy-in aspect is mostly what
that complaint is about. That the site is quite slow obviously
doesn't help, but that's probably the second or third complaint.
Our devs *hate* editing through the web, it seems (I know of options
like external editor, etc, which I tried to support, but I didn't
have much success with), and users seem more comfortable editing
wiki-style rather than through a workflow (even though I would argue
that our documents deserve some review and workflow).
The most significant problem is there are no Plone admins in our
community other than myself (and I only admin the MapServer website
-- no other Plone sites anymore) who've stepped up to help with the
site. Plone's failure for the MapServer website is entirely
manpower related, not software. The software's fine, it just takes
a lot of expertise and time to polish into what is needed to serve
the project. That investment, while significant, could provide a
lot of leverage and power to the project, but a single volunteer
admin doesn't really have the capacity. MapServer wants some
customizations, but doesn't really want to pay for them in blood,
sweat, and tears. Copying another project's infrastructure (as the
RFC proposes to do) that closely maps what we want to do seems to be
the easiest way to get some leverage without investing a whole lot
(and hopefully garner one or two capable admins in the process).
If I offended anyone with my venting, I'm sorry. A user could go
venting off saying MapServer is a bitch at X, Y, or Z, and while it
might be valid criticism in some instances, it would still sting
given the amount of effort that's gone into the thing.
PS, I just subscribed to the evangelism list if you wish to continue
the discussion there.
Evangelism mailing list