action #64541
open
Issues with DNS resolving in QEMU
Added by favogt about 4 years ago.
Updated over 3 years ago.
Description
Currently many tests break because DNS requests fail. Most of the failures are regarding "production.cloudflare.docker.com" when doing a docker/podman pull, but there are also other network related issues.
When trying often enough, the lookup starts working at some point.
I recorded a tcpdump trace of a broken DNS lookup done by podman pull, it is attached.
I tried to replicate the DNS request with "dig +notcp +noadflag +nocookie @10.0.2.3 production.cloudflare.docker.com", but it succeeded. After that, podman started to work as well though.
Files
Just tried the dig command again in the same SUT, it failed now for some runs, but then started working again.
So not a docker/podman issue.
- Priority changed from Normal to Low
- Target version set to Ready
- Status changed from New to Feedback
- Assignee set to okurz
@fvogt seems I have overlooked this issue for long. Is this the same as the one discussed in the weekly "openSUSE release management meeting" where mainly Lubos is providing minutes? IIRC the last mentions were that it's either not severe anymore or fixed. Can you provide a current status?
- Status changed from Feedback to Blocked
okurz wrote:
@fvogt seems I have overlooked this issue for long. Is this the same as the one discussed in the weekly "openSUSE release management meeting" where mainly Lubos is providing minutes?
Yep, but the issue changed in nature: After switching from the old DNS (.4) to the new DNS (.61) plain DNS requests worked again, but DNSSEC broke. This only affects the support server tests now.
IIRC the last mentions were that it's either not severe anymore or fixed. Can you provide a current status?
It's still broken in the same way. As a workaround, ariel is using 8.8.8.8 as nameserver and not the infra provided one(s).
The infra ticket is https://infra.nue.suse.com/SelfService/Display.html?id=174934 and the latest state is unfortunately that the issue isn't really well understood. It seems like the error is also intermittent and influenced by caching.
- Assignee changed from okurz to favogt
- Target version changed from Ready to future
ok, thanks. As you opened the EngInfra ticket and will be informed about any updates I hope you don't mind if I assign this ticket to you then because you can react and update accordingly whenever there is an update. As there seems to be a sufficient workaround in place I assume the SUSE QE Tools team does not need to actively followup with this ticket, hence setting "future" target version.
Also available in: Atom
PDF