-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256
Hello!
I've setup initial quick-fedora-mirror using fedora-buffet master module and MODULE=(fedora-enchilada fedora-epel) in quick-fedora-mirror.conf.
It's seems like there's never ending rsync transfer (cmiiw). My last log using verbose=3 output is :
file has vanished: "/fedora/linux/updates/24/i386/drpms/distribution-gpg-keys-1.4-1.fc24_1. 9-1.fc24.noarch.drpm" (in fedora-buffet) file has vanished: "/fedora/linux/updates/24/i386/drpms/distribution-gpg-keys-1.8-1.fc24_1. 9-1.fc24.noarch.drpm" (in fedora-buffet) file has vanished: "/fedora/linux/updates/24/i386/drpms/distribution-gpg-keys-copr-1.8-1.fc 24_1.9-1.fc24.noarch.drpm" (in fedora-buffet) file has vanished: "/fedora/linux/updates/24/i386/drpms/esniper-2.31.0-5.fc24_2.32.0-1.fc24 .i686.drpm" (in fedora-buffet)
Nothing changes, but the network transfer still exists through dl{0-4}.fedoraproject.org. Is it normal? Then should I break the existing initial run and put it on crontab like old mechanism? Or just moving it to systemd background process (since it's use zsh shell).
My fedora linux repo url is http://rndc.xtremenitro.org/fedora/linux My fedora epel repo url is http://rndc.xtremenitro.org/fedora/epel
Need advice.
Thanks in advance
"DBA" == Dewangga Bachrul Alam dewanggaba@xtremenitro.org writes:
DBA> It's seems like there's never ending rsync transfer (cmiiw). My DBA> last log using verbose=3 output is :
Well, if you don't have that data locally already, you're going to have to download about 2.4TB of files:
pubmirror1:/srv/mirror/pub/fedora-buffet❯ du -sk fedora epel 2224437692 fedora 175050724 epel
I should probably enhance the documentation a bit to better cover the initial download case, or even add an --initial option that doesn't do anything except run the proper rsync call once.
DBA> file has vanished: DBA> "/fedora/linux/updates/24/i386/drpms/distribution-gpg-keys-1.4-1.fc24_1. DBA> 9-1.fc24.noarch.drpm" (in fedora-buffet)
That just means that the repository has changed since you started downloading. fedora-enchilada is quite active so this is not at all surprising.
DBA> Nothing changes, but the network transfer still exists through DBA> dl{0-4}.fedoraproject.org. Is it normal?
Well, at that verbosity level it's not going to show you regular rsync output, just the rsync stderr and some extra debug info about what it's doing. If you look in /tmp/quick-mirror* you should see some rsync-out-*.log files; tail the most recent one to see things happening. Or run find -mmin -1 over your repository.
DBA> Then should I break the existing initial run and put it on crontab DBA> like old mechanism?
It's safe to stop it. Even though rsync is running with --delay-updates (which normally isn't safe to restart), quick-fedora-mirror will find that partial download and move it into place because that's better than endlessly accumulating things .~tmp~ directories.
You can run the script however you like; it doesn't particularly care. It will only allow itself to be run once but will warn if the lock has been held for too long. It will produce output which is intended for the administrator, so if you want to see that a cron job is good. I recommend verbosity level 1 for that case, which will be silent (and therefore not generate a cron email) if there's nothing to download and will otherwise generate timing and show you any errors.
I run it every ten minutes, but that's probably overkill.
- J<
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256
Hello!
Thanks for your response, initial repo was done and everything is synced now. Scheduled every 4 hours.
OOT, initial repo was generated in 2 days. Phew! verbose=1 looks fine to see the `--stats` than verbose=3.
My mirror is up and ready to serve the traffic.
Thanks
On 03/25/2017 01:13 AM, Jason L Tibbitts III wrote:
"DBA" == Dewangga Bachrul Alam dewanggaba@xtremenitro.org writes:
DBA> It's seems like there's never ending rsync transfer (cmiiw). My DBA> last log using verbose=3 output is :
Well, if you don't have that data locally already, you're going to have to download about 2.4TB of files:
pubmirror1:/srv/mirror/pub/fedora-buffet❯ du -sk fedora epel 2224437692 fedora 175050724 epel
I should probably enhance the documentation a bit to better cover the initial download case, or even add an --initial option that doesn't do anything except run the proper rsync call once.
DBA> file has vanished: DBA> "/fedora/linux/updates/24/i386/drpms/distribution-gpg-keys-1.4-1.fc24_
1.
DBA> 9-1.fc24.noarch.drpm" (in fedora-buffet)
That just means that the repository has changed since you started downloading. fedora-enchilada is quite active so this is not at all surprising.
DBA> Nothing changes, but the network transfer still exists through DBA> dl{0-4}.fedoraproject.org. Is it normal?
Well, at that verbosity level it's not going to show you regular rsync output, just the rsync stderr and some extra debug info about what it's doing. If you look in /tmp/quick-mirror* you should see some rsync-out-*.log files; tail the most recent one to see things happening. Or run find -mmin -1 over your repository.
DBA> Then should I break the existing initial run and put it on crontab DBA> like old mechanism?
It's safe to stop it. Even though rsync is running with --delay-updates (which normally isn't safe to restart), quick-fedora-mirror will find that partial download and move it into place because that's better than endlessly accumulating things .~tmp~ directories.
You can run the script however you like; it doesn't particularly care. It will only allow itself to be run once but will warn if the lock has been held for too long. It will produce output which is intended for the administrator, so if you want to see that a cron job is good. I recommend verbosity level 1 for that case, which will be silent (and therefore not generate a cron email) if there's nothing to download and will otherwise generate timing and show you any errors.
I run it every ten minutes, but that's probably overkill.
- J<
mirror-admin@lists.fedoraproject.org
-
Dewangga Bachrul Alam -
Jason L Tibbitts III