Hi Fritz, As you know, running *zypper dup* has various steps. High level this is my understanding: 1. Refresh of the local repositories (Internet) 2. Pulling down RPMs (Internet) 3. Applying the RPMs (local) I find the refreshing to be somewhat static. The next step of pulling down RPMs is dependent on one's Internet connection along with the demand on the mirror. Just in case, here's more information on the mirrors - https://en.opensuse.org/openSUSE:Tumbleweed_mirrors The final step is applying the RPMs. It is my limited understanding that applying each RPM is treated as an individual transaction: all-or-nothing. The serialization is great should there be a system interruption (power failure) but it also has a performance hit. If you want to have *zypper* create use a single RPM transaction (you have a UPS or a laptop), you can set the following environment variable: ZYPP_SINGLE_RPMTRANS=1. I also have set ZYPP_MEDIANETWORK=1. Here are the definitions of each from the mother ship[1]: * |ZYPP_MEDIANETWORK=1|Turn on the new multithreaded backend (experimental) * |ZYPP_SINGLE_RPMTRANS=1|Enable alternative and !!!experimental!!! commit strategy where all rpm operations are executed in a single rpm transaction, which results in much faster commits. I hope it helps! Cheers, ::: References ::: [1] - Zypper environment variables https://doc.opensuse.org/projects/libzypp/HEAD/zypp-envars.html --- Pablo Sanchez - Blueoak Database Engineering On 2023-03-07 12:52, Fritz Hudnut wrote:
Emanuel:
Thanks for the post back on it. Several months back I did have a forum thread going into what I think is the same continuing issue, and "we" did look at journalctl logs . . . can't remember the conclusions. Usually on the forums the gurus say, "You don't know what you are doing," and that is the extent of it. At that time it was, "You need to run as root, and in screen -L" . . . so since then I run a more or less box stock TW system, maintained as zypper determines . . . as root and in screen, etc.
Previously TW was having huuuugggeee number of packages with every zypper, which also took a very long time to run through; in the recent months the package numbers are down, but handling them seems to require considerable "time" in TW . . . . Manjaro is another rolling distro that I have, and some weeks 1 package, but other times 300 -454 packages, but pacman seems to be able to "zyp" through them asap . . . .
I think back then there was some mention of using the "time" modifier on the zypper dup move, but not all of the systems I have seem to be able to do that "time" feature, so I can't show "head to head" evidence of the comparative slowness of TW. I can only "feel" it. : - 0
F
On Tue, Mar 7, 2023 at 9:21 AM Emanuel Castelo <emanuel.castelo@icloud.com> wrote:
There is something else going on besides the process of ’zypper dup’ , I would check journalctl and install log.
> On Mar 6, 2023, at 12:50 PM, Fritz Hudnut <non.space.1@gmail.com> wrote: > > et al: > > Just reporting a "seat o the pants" impression that TW is very slow in running zypper dup . . . in comparison to my Leap 15.5, and my two Gecko rolling editions . . . not bringing in the Manjaro, the Bookworm, and the Sid editions on my '12 Mac Pro. > > Only difference is TW is the "grub handler" system, but today, running "195" packages and #update-bootloader . . . the machine was running for approx. 1.5 hours????? > > The difference in speed is "palpable" . . . in terms of running time consumed vs the Gecko systems. > > F