Hi,
From time to time I see different errors for dnf-makecache in the logs. e.g. „Errors during downloading metadata“, „Couldn't connect to server for“, „Timeout was reached“, „Couldn't resolve host name " or „Operation too slow“.
In all cases, a restart of the service solves the „problem“ and in other cases the next run dose it.
Beside having the errors, our current service monitoring triggers in most cases a warning message, and depending on the solution recovers too, which makes me wonder what would be the best practice for such situation.
a) do I need dnf-makecache? Will it speed up things so much and makes package management much more easy, so having some errors is o.k.?
b) adjust the config for dnf-makecache, so it will trigger less errors?
c) adjust our monitoring to ignore some errors or be more soft with triggering warnings?
C is the least preferred option right now.
What do you think, how do you handle dnf-makecache?
Thanks for feedback and suggestions. Götz
Am 08.11.21 um 09:59 schrieb Götz Reinicke:
Hi,
From time to time I see different errors for dnf-makecache in the logs. e.g. „Errors during downloading metadata“, „Couldn't connect to server for“, „Timeout was reached“, „Couldn't resolve host name " or „Operation too slow“.
In all cases, a restart of the service solves the „problem“ and in other cases the next run dose it.
Beside having the errors, our current service monitoring triggers in most cases a warning message, and depending on the solution recovers too, which makes me wonder what would be the best practice for such situation.
a) do I need dnf-makecache? Will it speed up things so much and makes package management much more easy, so having some errors is o.k.?
b) adjust the config for dnf-makecache, so it will trigger less errors?
c) adjust our monitoring to ignore some errors or be more soft with triggering warnings?
C is the least preferred option right now.
What do you think, how do you handle dnf-makecache?
It depends on your needs. We are okay with; stopping and disabling dnf-makecache.timer for prod systems.
We have dnf-automatic enabled. So, the cache is rebuilded anyway ...
-- Leon
Hi and thanks.
Am 08.11.2021 um 13:09 schrieb Leon Fauster via CentOS centos@centos.org:
Am 08.11.21 um 09:59 schrieb Götz Reinicke:
Hi, From time to time I see different errors for dnf-makecache in the logs. e.g. „Errors during downloading metadata“, „Couldn't connect to server for“, „Timeout was reached“, „Couldn't resolve host name " or „Operation too slow“. In all cases, a restart of the service solves the „problem“ and in other cases the next run dose it. Beside having the errors, our current service monitoring triggers in most cases a warning message, and depending on the solution recovers too, which makes me wonder what would be the best practice for such situation. a) do I need dnf-makecache? Will it speed up things so much and makes package management much more easy, so having some errors is o.k.? b) adjust the config for dnf-makecache, so it will trigger less errors? c) adjust our monitoring to ignore some errors or be more soft with triggering warnings? C is the least preferred option right now. What do you think, how do you handle dnf-makecache?
It depends on your needs. We are okay with; stopping
hm, I don't see any real benefit in our environment. If I do an update, waiting few seconds more or less for the repo data to be downloaded, dose not matter.
and disabling dnf-makecache.timer for prod systems.
We have dnf-automatic enabled. So, the cache is rebuilded anyway …
I see.
/Götz
-
Götz Reinicke -
Leon Fauster