249
edits
Changes
Incident Report: 20230109 (view source)
Revision as of 21:05, 11 January 2023
, 21:05, 11 January 2023no edit summary
|brief=ITS replaced some networking gear and broke some routes. Sad panda.
|brief=ITS replaced some networking gear and broke some routes. Sad panda.
|severity=High
|severity=High
|impact=Medium (anything URY-related unavailable for around ~8 hours)
|impact=High (anything URY-related unavailable for around ~8 hours)
|start=2023-01-09 09:30
|start=2023-01-09 09:30
|end=2023-01-09 17:19
|end=2023-01-09 17:19
Early in the morning of the first day of the first week of term, IT Services (as we later found out) replaced some networking equipment in the Vanbrugh area, and may have incorrectly set some static routes. The first we found out about this is at 9:30, when all of our monitoring pinged that URY had dropped off the face of the internet.
Early in the morning of the first day of the first week of term, IT Services (as we later found out) replaced some networking equipment in the Vanbrugh area, and may have incorrectly set some static routes. The first we found out about this is at 9:30, when all of our monitoring pinged that URY had dropped off the face of the internet.
Much scrambling and some reboots later, we narrowed down the state of the network to:
Much scrambling and some (ultimately futile) reboots later, we narrowed down the state of the network to:
* Anything with a 144.32.64.160/27 IP (so all of URY) could send packets ''out'' of URY, but no packets would make it ''in''.
* Anything with a 144.32.64.160/27 IP (so all of URY) could send packets ''out'' of URY, but no packets would make it ''in''.
:* With the exception of uryfw0 which has a separate IP (144.32.109.64).
:* With the exception of uryfw0 which has a separate IP (144.32.109.64).
::* You really don't want to know how... no, seriously, it's horrible...
::* You really don't want to know how... no, seriously, it's horrible...
An ITS ticket was filed at 12:12, and at around 17:14 service was restored.
== Lessons Learned ==
== Lessons Learned ==
* We need to be more happy to assume, if we haven't touched the network and something's happened, it could be an ITS issue. Look at things like traceroutes earlier.
* ITS don't really know about us - like they'd have just assumed uryfw0 is just 144.32.109.64 and not a gateway for all of 144.32.64.160/27, so be happy to remind them of this.
* Phone ITS earlier if it's a big problem.
* Don't get distracted by things that are merely the result of the problem - i.e. our DNS is often external to URY (i.e. wogan or ITS nameservers), so it can't resolve DNS. But it doesn't mean the key problem is DNS - the key problem is that no traffic reached us. What didn't help was the confusion about guest PC being able to access the internet - this is because it has a web proxy (because IRN).
[[Category:Incident Reports]]
[[Category:Incident Reports]]