SliTaz Forum
You are not logged in.
Pages: 1
Hello guys,
Since yesterday the $ tazpkg upgrade returns
You have 323 available upgrades (0 blocked)
351 installed packages scanned in 8s
Do you wish to install them now? (y/N)
This just does not look right. Any ideas?
Please let me know if any additional info would help.
Thank you,
-Z.
Offline
Hi Zann,
It's not mistake. Really, package “factory” cook.slitaz.org working since March, 6 without stopping, rebuilding all the SliTaz packages. If package rebuild, tazpkg offer you to update this package — even if nothing changed inside it.
Offline
Thank you Aleksej,
That explains the large number of available updates, but a few things that I am seeing still don't make a lot of sense.
One of them is large amounts of "Checksum error for package-name.tazpkg" during a single [c]$ tazpkg upgrade[/c] session. There seem to be two kinds of these: reported before and after a package has been downloaded. In the former case they are preceded by "md5sum: can't open 'package-name.tazpkg': No such file or directory" and followed by "rm: can't remove 'package-name.tazpkg': No such file or directory". In the latter case, they are followed by "Please wait until mirror synchronization is complete and try again".
Another is that the above checksum errors appear to trigger "Recharging repository "Main" with inconsistent result: sometimes it is "Repository "Main" is up to date", sometimes it is "Last database is ready to use", and sometimes "## new packages on the mirror" - only seconds apart!
Third is that the very same packages that were just updated are offered again and again, only minutes (if even that) after the previous seemingly successful installation. For example, slitaz-configs version 275 was installed some 3-4 times already tonight, but the next [c]$ tazpkg upgrade[/c] offers it again as upgrade from 273. Maybe I should have rebooted after each upgrade session?
Fourth is that [c]$ tazpkg upgrade[/c] issued 4-5 times in a row returns inconsistent results. For example, it could be "System is up-to-date..." the first two times, then "You have 14 available upgrades (0 blocked)", and then "System is up-to-date..." again - even if the 14 updates just offered were actually dismissed with "N" - why are they not offered again?
I looked at the wok and cooker, but could not see any correlations. Could it be that something is happening with the http://mirror.slitaz.org/packages/cooking used by my TazPkg?
I guess I worry too much and should simply wait till things calm down a little 
-Z.
Offline
Hi Zann,
There is the few minor issues normally we can live with. But when they are combined all together it's becomes a hell...
I advice you never update these days until mirrors are stabilized.
If you still need to update some packages now — please never use mirror.slitaz.org as your mirror. You can change your settings by editing [c]/var/lib/tazpkg/mirror[/c] configuration file, or in the GUI you can start by executing next command:
[c]tazbox su tazpanel pkgs#admin[/c]
Just click on your new mirror, and that's all.
I just tested quickly our mirrors, and next seems to usable right now:
[c]ftp://ftp.br.debian.org/slitaz/
ftp://opensuse.c3sl.ufpr.br/slitaz/
ftp://slitaz.c3sl.ufpr.br/slitaz/
http://slitaz.c3sl.ufpr.br/[/c]
Sure make updates at your own risk.
Offline
And now some details, because I see you're interested in them, not only in the fast solution. Also, other developers may give some feedback here.
Mirrors are not intended to fast delivery updates. SliTaz project uses several different mirrors. Some of them updates frequently — 3–4 times a day (I guess), when other updates once a several days. It is minor issue — you get new packages anyway, but maybe day or two later (at worst).
Now, when all the packages are still rebuilds about up to a week — it's too much work to our mirrors. They need to update all the packages — it's need a lot of traffic, it's slow, and consume much time.
Mirror update is not atomic. When you access specific mirror, some files may be updated while some other files still old. It is big issue, but we can't solve it now, I guess. Only I can advice to wait evening and try again in case of synchronization issues. I have some ideas, but far from the realization.
Issue with the “checksum error” is caused this non-atomic mirror update: packages database may be new while some packages itself may be old. Or packages DB may be old while some packages are new. Also, you get this error because I wrote not very good code in the [c]tazpkg[/c]: normally it installs packages from a cache. So, if it's a package you need is existed in the cache — tazpkg will install it, otherwise tazpkg will download it to a cache and then will install. Problem is — tazpkg not know if the package in the cache is downloaded right now or is downloaded early. If checksum of the cached package is not matched checksum defined in the packages DB — tazpkg fires a warning “checksum error”, removes the package from cache, updates the packages DB, downloads package to cache one more time; if error repeats it gives up with the message advising you to wait until mirror synchronized. Tazpkg may be a bit clever — if the package was downloaded to a cache a time ago — we can believe it's outdated and silently reload them without any warnings and packages DB reloading. Sorry, I'll work with this issue later after few days.
mirror.slitaz.org is not a mirror, — it is a multiplexer.
[c]LC_ALL=C wget -O/dev/null http://mirror.slitaz.org/packages/5.0/IDs
--2016-03-12 15:15:52-- http://mirror.slitaz.org/packages/5.0/IDs
Resolving mirror.slitaz.org (mirror.slitaz.org)... 37.187.4.13, 188.165.229.172, 212.83.135.187
Connecting to mirror.slitaz.org (mirror.slitaz.org)|37.187.4.13|:80... connected.
HTTP request sent, awaiting response... 301 Moved Permanently
Location: http://mirror.switch.ch/ftp/mirror/slitaz/packages/5.0/IDs [following]
--2016-03-12 15:15:52-- http://mirror.switch.ch/ftp/mirror/slitaz/packages/5.0/IDs
Resolving mirror.switch.ch (mirror.switch.ch)... 130.59.10.36, 2001:620:0:8::20
Connecting to mirror.switch.ch (mirror.switch.ch)|130.59.10.36|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 44 [text/plain]
Saving to: '/dev/null'
/dev/null 100%[======================>] 44 --.-KB/s in 0s
2016-03-12 15:15:52 (1.31 MB/s) - '/dev/null' saved [44/44][/c]
I tested it in the loop 10 times, and seems I wrong now. mirror.slitaz.org was the multiplexer — it was redirect query to one of the three random mirrors — it's not an issue when all that three mirrors are updated both. But it's a big issue if some of that mirrors are updated while other of them — are not. You tried to update “tazpkg” package, and few times you got new and old versions in random order.
Offline
Aleksej,
Thank you so much for such an elaborate reply. I think I am getting a pretty good idea about what might be happening already. And now I know where to look at for more clues too!
Thanks again,
-Z.
Offline
Pages: 1
[ Generated in 0.030 seconds, 7 queries executed - Memory usage: 1.55 MiB (Peak: 1.77 MiB) ]