[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Podman 3.0 and Debian bullseye



On Sunday, 31 January 2021 7:29:16 PM AEDT Shengjing Zhu wrote:
> I do run a private registry with my CI, many images are pushed and
> deleted daliy. But I haven't noticed disk space leaks.

Upstream have many bugs about that... Here are just a few (I've seen more
of those...):

  https://github.com/distribution/distribution/issues/1844
  https://github.com/distribution/distribution/issues/2190

Even with "docker-registry garbage-collect" some images are never deleted...

Also have a look at the following project (there are more but this one works)
- it have more relevant upstream bugs referenced:

  https://github.com/mortensteenrasmussen/docker-registry-manifest-cleanup

I'm not sure about the following one:

  https://github.com/TranceMaster86/docker-reg-gc


> > IMHO just saving built images to network share is the best, easiest, most
> > reliable way of deploying local images. It is the best to avoid Docker
> > registry whenever possible.
> 
> However only the registry is able to dedup the image layers.

But Docker Registry is also the single point of failure.
In otherwise redundant cluster such as Nomad one can strive to avoid
single point of failure. Network share on clustered file systems
(e.g. MooseFS, LizardFS, GfarmFS, etc.) can be more reliable than
a NFS share from single server or a single instance of Docker Registry.

-- 
Kind regards,
 Dmitry Smirnov
 GPG key : 4096R/52B6BBD953968D1B

---

The trouble with our times is that the future is not what it used to be.
    -- Paul Valery

---

"A closer look at U.S. deaths due to COVID-19"
    2020-11-26, The Johns Hopkins News-Letter
    -- https://notthebee.com/article/a-few-days-ago-johns-hopkins-published-a-study-saying-corona-is-nbd-they-then-deleted-it-read-it-here-in-its-entirety
    -- https://web.archive.org/web/20201126223119/https://www.jhunewsletter.com/article/2020/11/a-closer-look-at-u-s-deaths-due-to-covid-19

Attachment: signature.asc
Description: This is a digitally signed message part.


Reply to: