Re: [HTCondor-users] zswap for jobs

[Bert DeKnuydt <Bert.DeKnuydt@xxxxxxxxxxxxxxxx>]

On 2026/04/08 12:10 pm, Emily Kooistra wrote:
> Hi,

Hi Emily,

> Does anyone have operational experience with running zswap on EPs?

Yes.  I've been running zswap for years, on almost everything, including EPs.  Here, most jobs use little or no
swap at all, but some just 'explode', often only temporarily. I've always had the (admittedly: subjective) feeling
that EPs tend to survive such events (as in remaining responsive) much better using zswap.  In fact, I see little
disadvantage in enabling it (as compared to zram).

> I was wondering what the compression rates people are getting for this?

No hard numbers; I doubt for real data it's more than 2.

> Right now there is no native support for limiting the amount of this in the cgroups (via htcondor) (https://www.kernel.org/doc/html/latest/ 
> admin-guide/mm/zswap.html), but even without this i would assume that it can give some nice gains with overprovisioning memory to some regard.
>
> Specifically given, with the "default" cgroup hard limits, we have right now quite a lot of "spare" ram that could be used for disk caches 
> or other reasons, but increasing the limits too much is also no ideal, so exploring possibilities on how to make more efficent use of the 
> resources wile also preventing running out of memory if memory intensive resources are running.

In my opinion, that's the advantage of zswap: the kernel manages the lot, there's no tuning needed (balancing between
zwap and disk-cache).

But curious to anyone else's experiences...

I recently read https://chrisdown.name/2026/03/24/zswap-vs-zram-when-to-use-what.html which points to some pro's and con's of
it. He also suggest to 'just switch it on' in most cases.

Cheers, B.