Over 5000 updates!

English Install/Boot/Login
21

Sure. However as already mentioned above upgrading several thousand packages is a non-issue. To my experience an upgrade failing due to high server load eventually succeeds when performed the day after.

Host erlangen gets a daily upgrade. Most of them are performed in less than one minute. Only some of them last longer:

erlangen:~ # journalctl -q -u dup -g Consumed|grep min
Jan 24 06:10:39 erlangen systemd[1]: dup.service: Consumed 1min 8.964s CPU time.
Feb 04 03:49:08 erlangen systemd[1]: dup.service: Consumed 1min 53.400s CPU time.
Feb 05 15:54:18 erlangen systemd[1]: dup.service: Consumed 1min 32.727s CPU time.
Feb 08 19:32:16 erlangen systemd[1]: dup.service: Consumed 1min 31.952s CPU time.
Feb 14 23:23:53 erlangen systemd[1]: dup.service: Consumed 1min 33.783s CPU time.
Feb 16 04:24:12 erlangen systemd[1]: dup.service: Consumed 1min 347ms CPU time.
Mar 07 18:23:11 erlangen systemd[1]: dup.service: Consumed 3min 16.117s CPU time.
Mar 17 03:23:04 erlangen systemd[1]: dup.service: Consumed 1min 16.042s CPU time.
Mar 19 14:24:48 erlangen systemd[1]: dup.service: Consumed 1min 4.490s CPU time.
Mar 23 03:44:46 erlangen systemd[1]: dup.service: Consumed 7min 18.557s CPU time.
erlangen:~ #
22

To summarize some of the great advices here:

  • The best distro for one needs to take into account the context. In this case in particular an openSUSE rolling release performs better with a fast connection to internet. To mitigate that:
    • Run zypper dup less often, once every week / two weeks;
    • Notice if a big snapshot is incoming, and cancel zypper dup, try again next day;
    • Prefer MicroOS where a fewer number of OS packages is required, while apps are updated as flatpaks less frequently

Totally anecdotal, but I see fewer regular rebuilds than it used to be. Most package updates include package source updates.

23

Jumping in since I was tagged, and also not a “new user” of TW . . . .

Rolling release is not a synonym for “rebuild the whole distribution every couple of weeks”.

This is the key point, what is “amusing” to me is that the number of packages to upgrade in TW of late would actually exceed the number of packages in a fresh install??? That is “over my head” . . . ???

As an “end user” and as mentioned by OP and others, perhaps @hui himself, the joy of TW is indeed “fresh horsies” . . . and the reason others, such as myself, prefer “rolling” distro is because there isn’t the need to do a total system upgrade, as in ubuntu, every 6 months. I have other “rolling distros” installed in various machines, and they don’t seem to need this “balloon upgrade” approach, that does become cumbersome, in either machine time, or mental attention . . . . Somebody else mentioned “curl errors” that would befuddle the process . . . .

Historically, going back perhaps ten years of using TW, there was not these large tsunami upgrades the way that it has become as of the last 2 or 3 years . . . ??? That is a decision that IMHO might be reviewed.

Case in point, on my Manjaro rolling edition yesterday, not booted for several weeks, I ran their version of zypper dup . . . it said, “running full system upgrade” and then cursor returned . . . “nothing to do.” ??? “Over my head” how they can offer, nothing to do to maintain my system in working order??

I’m booted in Leap 15.5 today . . . I think package upgrade notification showed me “two upgrades” are waiting to be run . . . such a low number, might not run it.

“Rolling release” does not mean “rebuild the system beyond what was installed every couple of months” . . . TW has been a traditional release that “rolled along” . . . what happened to that approach to maintaining a fresh system???

2 Likes
24

This is when the admin/user fails to properly maintain his system and installs a lot of useless stuff to his system without removing unneeded stuff…

25

@non_space hopefully users will follow the Factory Mailing list, especially this weeks ‘Review’…

Tumbleweed IMO really isn’t a rolling release, it is a new distribution release with every snapshot running the latest foo :wink:

1 Like
26

I also like to DUP daily, Fudokai, but have learned to wait a few days if the latest snapshot shows 2,000 or 3,000+ packages. The main download servers are less likely to be bogged down then. (Or you could switch manually to another mirror, as hui suggested.)

I’m sorry that you had this experience, and know that I also would have felt frustrated if I’d needed a full day to DUP a single laptop. You have my sympathy.

2 Likes
27

Was “only” 1’800 packs for me but still took 6 hours a day after the release. Same problem with hitting “r” every now and then.
If the whole system gets rebuild I guess that is normal. I was also surprised as I never had such a not big but time consuming update :sweat_smile:

2 Likes
28

I killed zypper two days in a row because it didn’t go ahead with the update, despite having the 100 megabit line the update reached a few kbits.

1 Like
29

Well, you have to anyway. This thing happens 1-2 times a year. IIRC it’s due to the upgrade of gcc to gcc13 .

1 Like
30

I use leap 15.4 as my main OS. I run Tumbleweed in a VM. I only use the TW vm for apps that won’t run in Leap because they are new software that need newer dependencies to run or compile or are unavailable in Leap. I just started the large TW update in the vm, turned it on, and went on to other things in leap 15.4.

31

Servers are back to business as usual:

Linux:~ # journalctl -b -u dup -g 'Start|folgende|Consumed'
Mar 24 19:14:18 Linux systemd[1]: Started Dist Upgrade.
Mar 24 19:14:52 Linux zypper[3025]: Die folgenden 4018 Pakete werden aktualisiert:
Mar 24 19:14:52 Linux zypper[3025]: Die folgenden 26 Schemata werden aktualisiert:
Mar 24 19:14:52 Linux zypper[3025]: Das folgende Produkt wird aktualisiert:
Mar 24 19:14:52 Linux zypper[3025]: Die folgenden 26 Pakete werden durch eine ältere Version ausgetauscht:
Mar 24 19:14:52 Linux zypper[3025]: Das folgende Paket wird die Architektur ändern:
Mar 24 19:14:52 Linux zypper[3025]: Die folgenden 66 NEUEN Pakete werden installiert:
Mar 24 19:14:52 Linux zypper[3025]: Das folgende NEUE Schema wird installiert:
Mar 24 19:14:52 Linux zypper[3025]: Die folgenden 5 Pakete werden GELÖSCHT:
Mar 24 19:14:52 Linux zypper[3025]: Das folgende Paket erfordert einen Systemneustart:
Mar 24 20:03:04 Linux zypper[3025]: Es werden Programme ausgeführt, die immer noch die durch kürzliche Upgrades gelöschten oder aktualisierten Dateien oder Bibliotheken verwenden. Starten Sie die Programme neu, um die Aktualisierungen zu nutzen. Mit 'zypper ps -s' erha>
Mar 24 20:03:04 Linux systemd[1]: dup.service: Consumed 18min 36.740s CPU time.
Linux:~ #

Downloaded some 2,52 GiB and upgraded some 4000 packages in 49 minutes. Large upgrades are a non-issue with download speed 27,49 Mbit/s. Machine stayed usable during the whole upgrade.

1 Like
32

@malcolmlewis

I think that is a correct assessment of the current TW philosophy . . . when I first got into TW it was a rolling release, like many of my others . . . 2 -300 packages upgraded every other week, all went through fast and easy, etc.

I’m not checking the web page that shows “TW” and “leap” tabs for accuracy, but I recall it says TW is “rolling” . . . but it should be changed to reflect your insight. The “caveat emptor” statement should be the “entire system is rebuilt every few months and will mean extremely large package changes for complete freshness of everything” . . . rather than “this is a rolling distribution” . . . . : - )))

33

Here is wikipedia’s definition of a rolling release:

TW is a rolling release. It does not require a complete rebuild every few months; it requires a complete rebuild when the gcc compiler undergoes a upgrade that breaks abi compatility. This is not a TW specific issue, and is not a rolling release issue; it is a timing issue. Leap and similar products typically release on a semiannual basis and any packages updated/added during that period are built on the initial tool chains. Other rolling products may choose to lock in their tool chains for arbitrary periods of time and forgo being leading-edge products for less frequent rebuilds.

The great thing about linux is it’s free, it has many different flavours, and you can choose a flavour that best meets your requirements.

This is the link to the Tumbleweed web page, and interestingly enough it does not use the term “rolling release”, instead it uses “continuously updated”.

34

My two cents: the total number of updates ,which is indeed a part of continuously updating, is balanced out for me by other things, such as the qa process and the structured delivery approach. And I am in Vietnam where mirror speeds are significantly lower. For me the pattern approach to updates brings along a lot of unwanted packages, in the last update at least 2000 files were related to texlive, which I never used (nor LaTex). If that concept was a bit more opt-in for users it would be great. So the solution for me: today I’ll be doing a fine-grained custom install (btw OpenSuse is the only distro that offers that much granularity), carefully selecting packages instead of patterns and moving on. The pluses by far still outweigh some minuses for me.

1 Like
35

Hoo boy! I seem to have opened up a real can of worms here.
Since I go back to installing (and reinstalling) SUSE (i.e that’s before openSUSE) from a heap of 3.5" disks and downloading updates over a 1200baud modem I’m not actually that disgruntled at the process of handling a rolling release over broadband - believe me, it’s a lot easier now!
I was more curious as to why so many applications got updated at the same time and just wanted to pass on what it felt like to those of us not on good broadband links :wink:
I gather it’s owing to an update to gcc - that’s all I really wanted to know - thanks!

3 Likes
36

You should be glad you’re not in China, otherwise you’d have a much worse experience with your network!!!

1 Like
37

I can relate to the OP’s surprise, and agree with others. As a quick example :

We have (at home) 3 laptops and 4 desktops running TW (been running TW for years). I run zipper dup on all machines daily.

My recent example:

On Mon (days are arbitrary, not exact) there’s 3 updates, Tue 2 updates, Wed 6 updates, Thur 2 updates, Fri 8 updates … then Sat BAM, almost 4,000 updates.

Sorry, shouldn’t be that many from one day to the next - the whole “it’s a rolling distro” is not a good argument. It’s a software release management scheduling oversight.

38

No, it is not. It is conscious decision to reduce ongoing maintenance burden for packagers. It is easier to make sure everything is built using the same version of compiler, library, … (pick your choice) once than play whack-a-mole by catching incompatibilities later.

39

Real data from infamous host erlangen:

erlangen:~ # journalctl -q -u dup -g Overall | cut -d ' ' -f 1-2,6-10
Feb 20 Overall download size: 228.8 MiB.
Feb 20 Overall download size: 19.8 MiB.
Feb 22 Overall download size: 60.6 MiB.
Mar 07 Overall download size: 1.22 GiB.
Mar 08 Overall download size: 127.7 MiB.
Mar 09 Overall download size: 83.8 MiB.
Mar 09 Overall download size: 87.7 MiB.
Mar 10 Overall download size: 148.7 MiB.
Mar 11 Overall download size: 158.2 MiB.
Mar 12 Overall download size: 295.1 MiB.
Mar 13 Overall download size: 393.3 MiB.
Mar 13 Overall download size: 190.1 MiB.
Mar 14 Overall download size: 235.4 MiB.
Mar 17 Overall download size: 661.7 MiB.
Mar 17 Overall download size: 31.7 MiB.
Mar 18 Overall download size: 892.8 MiB.
Mar 19 Overall download size: 17.3 MiB.
Mar 19 Overall download size: 337.2 MiB.
Mar 20 Overall download size: 112.9 MiB.
Mar 20 Overall download size: 80.8 MiB.
Mar 22 Overall download size: 109.3 MiB.
Mar 23 Overall download size: 3.47 GiB.
Mar 24 Overall download size: 152.2 MiB.
Mar 24 Overall download size: 293.0 MiB.
Mar 25 Overall download size: 5.4 MiB.
Mar 25 Overall download size: 682.3 KiB.
Mar 26 Overall download size: 772.9 MiB.
Mar 27 Overall download size: 221.3 MiB.
Mar 28 Overall download size: 181.8 MiB.
Mar 29 Overall download size: 100.0 MiB.
Mar 30 Overall download size: 32.5 MiB.
Mar 30 Overall download size: 506.1 MiB.
Apr 01 Overall download size: 138.1 MiB.
erlangen:~ #

Download size varies from day to day. However even large upgrades as on March 23 with 3.47 GiB run smoothly and unattended in the background. Users won’t even notice what’s going on. To my experience upgrading Tumbleweed is a non-issue.

With many machines to upgrade you may download once and share the download.

40

How do you actually deal with reboots?
After each zypper dup I usualy get the warning

There are running programs which still use files and libraries deleted or updated by recent upgrades. They should be restarted to benefit from the latest updates. Run 'zypper ps -s' to list these programs.

If zypper ps -s is telling me that core libraries got touched, I usually do a reboot. If the output only shows processes running under my own account, I do a logout/logon. If it’s including root, then I also do a reboot.

But that’s getting a bit annoying, so I was wondering how you deal with this situation.