HN Explorer

19 comments

[ 3.2 ms ] story [ 27.7 ms ] thread
java-man
Ah, that's why my builds were failing...
inglor_cz
Mine, too. I had a sleepless night, so around 4:30 am CEST I started attacking a certain problem in an Android app. Wham!
hgschmie
It's actually repo.apache.maven.org that is failing.

The repo.maven.apache.org is a CNAME for that.

foolfoolz
it’s been intermittent this afternoon, my recent builds fixed
quyleanh
I faced this error one time when building RISC-V.

I don't remember exactly the error but I have to fix by creating a script to manually download dependencies.

nikanj
It’s always DNS
pletnes
Except when it’s a compiler bug.
papandada
For a certain era for me, it was always "kerberos double hop"
javajosh
No sometimes its BGP, the stuff of nightmares.
nick__m
Internet-scale BGP is a nightmare, and it's a miracle that it works so well. But what makes BGP so nightmarish in a private enterprise network? As a software engineer who pivoted to infrastructure, I find that BGP is a lot better than the series of proprietary link-level SLAs that it replaced.
zamadatix
I've always been partial to IS-IS for the IGP but I wouldn't say BGP is necessarily bad. Perhaps too many knobs and not exactly as speedy to react as you'd want for an IGP in all but basic topologies but both of those can be fixed by staying focused on tweaking a subset that makes sense. Also I think the whole iBGP vs eBGP thing drives people nuts.
dnablob
this

IGPs (IS-IS, OSPF, EIGRP) and BGP solve different problems and networks often need both. Some niche networks get away with using only BGP (massive datacenters), and some get away with using only an IGP (smaller networks), and some don't even need that (small networks).

jimnotgym
I don't really do IT support any more, but when 'the internet was broken' at work last week my first reaction was to open cmd and nslookup a week known site! On corporate networks a missing dns server has been a major cause of issues.