OK let's look at it another way . . . .
I have a problem.
"Show me the problem."
Here it is - no it's not here any more.
"I can't fix it if it isn't broken."
These kinds of things are a nightmare to debug, and worse for tech support who are often nowhere near the servers themselves. When a problem is
persistent (Oh, I can see the problem, let me try to fix it) or
consistent (whenever I click this link, it does it) it can be fixed. But if it comes and goes without apparent cause, it's really tough.
Let me offer one bit of advice, don't know if it's helpful or not. When these things come up, we put on our Sherlock Holmes caps - eliminate the impossible, and what remains must be the truth (or as close as we'll get to it.)
that was the error on the blog
It may be entirely possible the "blog" is doing something that is putting an unusual strain on the server. I have seen poorly coded applications cause an error: "the mySQL server has gone away." I would suspect, in this shared hosting environment, this blog is exceeding some limit or causing the mySQL server to crash. It comes back in a few minutes because it starts up again.
This is, of course, all speculation, but that's the first place I'd look. If your condition is consistent - "only when I do THIS" - it lends more credibility to this idea. If not, maybe it's the overall app, or maybe as lammert says, shared hosting can be extremely unreliable as you've no idea what any of the other 600 or so domains on this server are doing. Maybe stepping up to better hosting will be the only permanent solution.