I don't know what mail client you use but it turned your message (or
at least the text/plain part) into one unreadable mess of a long line.
Argl... thought I'd set it to plain text; clearly not. Sorry.
What follows would be the actually-readable version of it...
---
Hello,
For the past month or so now, we've been observing severe issues
affecting our availability to North-West US coast AT&T users when
reaching us through Tata communications.
It seems to occur daily, from roughly 4:40 AM UTC and until 7:30 AM UTC,
and connections get entirely dropped on their way from AT&T to Tata.
The issue might be more extensive geographically and or timespan-wise
but that's the specifics we've been able to confirm ourselves.
Using global traceroute (and specifically probe 3670, AT&T Sacramento),
the issue seems to revolve around 12.122.114.5 [AS7018]. Here's an
example traceroute of what's happening:
traceroute to 203.175.174.246 (203.175.174.246) 64 hops max, 76 byte
packets
1 10.0.0.1 (10.0.0.1) [AS???] 8.345 ms 2.064 ms 2.01 ms
2 68.78.72.22 (68.78.72.22) [AS7018] 2.755 ms 2.5789999999999997 ms
3.197 ms
3 * * *
4 71.147.199.98 (71.147.199.98) [AS7018] 20.544 ms 20.274 ms 20.285 ms
5 12.122.160.166 (12.122.160.166) [AS7018] 30.043 ms 28.155 ms 26.833
ms
7 12.122.114.5 (12.122.114.5) [AS7018] 25.962 ms 26.106 ms 26.115 ms
8 * * *
9 * * *
10 * * *
11 * * *
12 * * *
255 * * *
Interestingly, the other way around (Tata->AT&T) works just fine, making
its way into AT&T's network through 192.205.37.69 this time.
We'd hoped things would fix themselves and moved traffic around via
other ASNs in the meantime, but it seems that the problem is either not
known or at a standstill.
Does anyone know something more or experience the same issue?
From some unsubstantiated claims I have been given it seems like the
issue would be on Tata's side but it's been nothing more than hearsay
and a few vague links so far (such as
https://mailman.nanog.org/pipermail/nanog/2022-January/217126.html), so
I was hoping for something more definitive, or hearing about others
encountering similar issues.
Thanks,
Tristan
PS: The target host in the example above is not ours, nor affiliated
with us. Please don't bother them about us, they are merely affected in
the same exact way as we are.
