Select Page

About 8PM last night my Dreamhost Fail post from yesterday must have found its way into somebody’s hands @ Dreamhost and they finally got around to addressing my emails… sort of.  I got the following response, a mix of helpful (but non-affecting) advice and a not-so-thinly-veiled insult thrown in for good measure:

My name is Jason and I’m the resident WordPress nerd around these parts.
I’m also one of the tech support managers. Your case was brought to my
attention by one of our developers who keeps an eye on Twitter and looked
into your issue briefly before asking me to dive into things. With the
delay in response, it’s understandable how frustrating it must be to deal
with this sort of laggy back and forth – so I’m sorry about that.

Now I’ve read over your analysis and have done some of my own. I’ve spot
checked your WordPress installs and have gone over the active plugins in
your wp_options table. After doing so, I’ve noticed that you are running
“brianslatestcomments.php” on a number of your installations. Now I know
that might seem pretty standard, but it can drive the load up on sites
quite a bit. What’s more, I don’t really know of that plugin receiving
an upgrade for over a year now. And since the WordPress back end has
made leaps and bounds in that time, I’m sure you can understand how
running it might be detrimental.

Also, WP-Cache was abandoned by its developer some time back and isn’t
really recommended at all. There’s a list of caching plugins that are
compatible and play nicely (for the most part) on our wiki:

It’s really suggested you upgrade that above all else as there may also
be security holes in the final version of WP-Cache. There have been a
rash of hacks where people have been on 2.7 and have all their plugins up
to date – and the only thing I see in common is WP-Cache.

When it comes down to it, I get what you’re saying – and I can see how
maddening trying to pinpoint this issue must be. But the reason you’re
getting 500 errors is because you’re running out of memory. This is not
because you’re running afoul of a process monitor. That would defeat the
whole purpose of putting you on a PS to begin with. In fact, we don’t
install that script on a Private Server without a request from the
server’s user. The complaints you were seeing came from shared server

The reason you’re running out of memory is because something is driving
up the load on your server. I’ve pointed out two possible culprits and I
suggest you deal with them before going on another ill-informed (but
understandably angry) tirade on your blog.

I want to continue to work with you on this. Please keep me posted on
how my suggestions work for you and I’ll do whatever I can to make sure
you get every last bit of assistance you might need.


Note that things started working again with no changes except that I accidentally/on-purpose upped the memory allocated to the server (and am now paying handsomely for it) at around 6PM.  Top now reveals the following:


Lots of apache tasks still (18) but no zombies.

I removed “Brian’s latest comments” and “WP-cache” from my install, for the record, with no discernable shift in memory consumption.  The former wasn’t actually being used anyway, as WP2.7 now has this feature built-in.

I think the real issue is failure to communicate — nobody’s heard anything specific from Dreamhost since this all began.  And given that this seems to be afflicting many other users clustering around the same time, who also aren’t listed on the “spacey” issue, I am loath to write this whole thing down to a couple of WordPress plugins not playing nicely.

I still believe that Dreamhost has a process manager that is not playing nicely.  But I will do some optimizations and report back here.

UPDATE:  Optimizations complete, now I have 4 bare bones WP sites and a single site with 11 plugins @ … a few hours after stripping everything and following “Jason’s” recommendations I’m actually further behind:


Thirty-two (32!) sleeping versions of Dreamhost’s apache2-argon-h tasks.

I use Dreamhost’s services (and pay more for them than I would at SliceHost) so that I don’t have to spend all day admining boxes.  For the past three days, though, I’ve been trying to debug their service for them.

It’s not going well.