Ensure atomicity for lastsync file

See merge request !794

pita strudl authored 8 months ago and pita strudl committed 8 months ago

We try to ensure an atomic operation of the lastsync file. This requires creating a tmp file which needs to be ignored.
This should take care of having empty lastsync files being served.
Possible cause is that the IO is stuck thus taking several seconds to write the lastsync timestamp.
This causes mirrors to download the empty file which causes checks to fail.

One year of mirrorlogs keeps a table of 5000 MB which is rather large as
we only show 7 days of logs. Keep it 6 months as maybe in the future
mirror operators are interested in older data.

Keep postgresql connections around for 5 minutes this avoids expensive
authentication requests.

certbot: Use ECDSA (P-256) certificates, not RSA

See merge request !844

certbot switched to ECDSA by default about two years ago, following
[recommended practices][1].

We are currently using RSA with 4096 bits, which is extremely slow to
sign. Using ECDSA should give us a nice speedup.

[1]: https://wiki.mozilla.org/Security/Server_Side_TLS#Intermediate_compatibility_.28recommended.29

nginx: Update SSL settings to current guidelines

See merge request !843

This reduces the session cache size and adds the
`DHE-RSA-CHACHA20-POLY1305` cipher.

dbscripts: Enable proxy_cache_background_update for mirrorauth

See merge request !842

This allows serving a stale response even to the request that triggers
an update. This should ensure all requests finish quickly.

With just `proxy_cache_use_stale updating`, the request that attempts to
update the cache waits for the response, while all other requests get to
use the stale response.

Currently archweb is badly overloaded and can take over half a minute to
respond. Pacman is not that patient and fails the download.

planet.archlinux.org redirects to https://archlinux.org/planet which
then redirects to https://archlinux.org/planet/. Skip one extra
redirect.

As we aren't the speediest at checking, this is fine.

This rate limits the endpoint which does things (ie. uwsgi). 10 requests
per second was already a lot, so 5 should be fine, realistically it can
go lower as we have a burst.

Update archmanweb to v1.11

See merge request !841

See: #600

Starting September 2024 the lower pricing on our grandfathered CX11
cloud server will go away. Thus, the price will be the same as CX22.

Carsme has previously been a Junior Package Maintainer, but the two
month period is now up and their changes all looked good.

Related to #579



Signed-off-by: Christian Heusel <christian@heusel.eu>

Missed this earlier for the upgrade to Loki 3.0.

Loki 3.0 requires this to be set when retention is enabled.

This is needed to upgrade to Loki 3.0.

[1] https://gitlab.archlinux.org/archlinux/nvchecker-poc

archwiki: Update to 1.42.1-2

See merge request !839

This contains fixes to the ArchLinux extension for MediaWiki 1.42.

Adapt to new Hetzner plans[1] for lower cost and better specs!

See merge request !836

CX11  -> CX22: Same price and better specs!
CX21  -> CX22: Cheaper price and same specs!
CX31  -> CX32: Cheaper price and better specs!
CPX11 -> CX22: Same price and similar spec! (for consistency)
CPX21 -> CX32: Cheaper price and better specs!
CPX41 -> CX42: Cheaper price and similar specs!

Some of the CX11 servers are not rescaled as they are on older, cheaper
grandfathered plans.

[1] https://www.hetzner.com/news/new-cx-plans/

This will used to increase the level of automation.

Signed-off-by: Christian Heusel <christian@heusel.eu>

He was onboarded as a Junior Packager but mistakenly the wrong group was
assigned to him.

Fixes: 27a2c9ac ("Onboard bertptrs as Junior Package Maintainer")
Signed-off-by: Christian Heusel <christian@heusel.eu>

Onboard anonfunc as junior package maintainer

See merge request !838

Related: #595