SP metadata cache keeps growing

[Peter Schober]

* Cantor, Scott <cantor.2 at osu.edu> [2020-02-27 20:33]:
> On 2/27/20, 2:22 PM, Peter Schober wrote:
> > Is that something you'd consider adding while xerces takes its time?
> 
> I don't think I could, I'm not in control of the thread until it returns.

OK

> > In the meantime I've amended the systemd wik page a bit:
> 
> If it doesn't actually wait that long I could obviously default the
> unit file to something longer. I don't know how well that work if
> there's a real problem, but I assume it's smart enough to know that
> if the process exited, it's not coming back up.

Obviously that would need some testing and it's unclear to me what a
good value would be. (30m just turned out not to be sufficient on a
test system with 2GB RAM and 2 cores, for loading out edugain-enabled
local feed.)

I'd be tempted to go with TimeoutStartSec=infinity, deactivating this
protection completely, but then loading huge metadata files in the
foreground -- outside of a background thread -- only hurts the very
first time, AFAIR, and people probably shouldn't be doing that in the
first place given that MDQ now exists...

-peter