On Mon, 1 Aug 2022 at 07:45, Dusty Mabe <dusty(a)dustymabe.com> wrote:
On 7/29/22 12:05, Peter Robinson wrote:
>> Looks like dnf makecache is uses a lot more memory, causing issues on
>> smaller systems/containers.
>>
>> F34:
>>
>> Metadata cache created.
>> 1.51user 0.15system 0:12.01elapsed 13%CPU (0avgtext+0avgdata
162440maxresident)k
>> 144inputs+56outputs (0major+46906minor)pagefaults 0swaps
>>
>>
>> F35:
>>
>> Metadata cache created.
>> 29.28user 2.15system 0:49.94elapsed 62%CPU (0avgtext+0avgdata
841704maxresident)k
>> 184160inputs+497320outputs (181major+425900minor)pagefaults 0swaps
>>
>> Is this a known issue?
>
> I've seen it on arm systems with 512Mb RAM which previously ran dnf
> (not just makecache) fine and now don't. There was a bug opened but
> the dnf team closed it.
Seems like this bug is related
https://bugzilla.redhat.com/show_bug.cgi?id=1907030
We are hitting this issue in Fedora CoreOS CI on VMs with 1G of RAM.
I wonder if this is one of those problems where microdnf needs to be used
until the full dnf rewrite in C++ is done. I remember something about
memory usage and dnf vs microdnf a while ago for smaller memory systems..
and with the general 'we need to double memory usage' every couple of
releases that applications have ... maybe 1Gb is no longer valid?
--
Stephen Smoogen, Red Hat Automotive
Let us be kind to one another, for most of us are fighting a hard battle.
-- Ian MacClaren