Incidents/2022-05-09 exim-bdat-errors

document status: draft

Summary

Incident metadata (see Incident Scorecard)

Incident ID	2022-05-09 exim-bdat-errors	Start	2022-05-04 01:28
Task	T307873	End	2022-05-09 16:40
People paged	0	Responder count	5
Coordinators	Keith Herron	Affected metrics/SLOs	No relevant SLOs exist, nor are there published metrics to quantify the impact.
Impact	About 14 thousand emails were rejected

Starting on 2022-05-04 some email from Google's mail server began being rejected with the log message, "503 BDAT command used when CHUNKING not advertised". These errors went unnoticed until 2022-05-09. After some investigation it was determined that disabling chunking support in Exim would mitigate the errors. During the time span of the incident about 14 thousand emails were rejected with a 503 error code.

Timeline

All times in UTC.

2022-05-04

• 01:28 Exim begins rejecting some email with 503s (Impact Begins)

2022-05-08

• 17:07 (bcampbell) Opens a ticket saying that multiple users have reported their emails being bounced with errors, https://phabricator.wikimedia.org/T307873

2022-05-09

• 8:25 (jbond) replies to ticket and begins investigation
• 13:55 (herron) Incident declared Herron becomes IC
• 14:00 (jhathaway) rolls back recently upgraded kernels, no effect
• 16:03 request to ITS to ask Google if anything has changed on their end
• 16:40 (jhathaway) chunking disabled in Exim, which successfully mitigates the incident, (Impact Ends)

Detection

The issue was first detected by users sending emails, https://phabricator.wikimedia.org/T307873. Though, the messages were rejected by Exim with a 503 error code. We do graph the number of bounced messages, but our alerting did not pick up these bounces, https://grafana.wikimedia.org/d/000000451/mail?orgId=1&from=1651536000000&to=1652227199000

Conclusions

Email monitoring of bounced messages does not account for all bounces.

What went well?

• We had a good group of SREs jump on the problem when the severity of the incident was understood.

What went poorly?

• The severity of the issue was not understood until five days after we first started seeing errors

Where did we get lucky?

• Faidon jumped into help with his extensive Exim and email knowledge even though he is no longer in a technical SRE role.

How many people were involved in the remediation?

• 4 SREs and 1 incident commander

Links to relevant documentation

• https://wikitech.wikimedia.org/wiki/Exim

Actionables

• Improve monitoring, https://phabricator.wikimedia.org/T309237

Scorecard

Incident Engagement™ ScoreCard

	Question	Answer (yes/no)	Notes
People	Were the people responding to this incident sufficiently different than the previous five incidents?	yes
	Were the people who responded prepared enough to respond effectively	yes
	Were fewer than five people paged?	yes
	Were pages routed to the correct sub-team(s)?	no
	Were pages routed to online (business hours) engineers?  Answer “no” if engineers were paged after business hours.	no
Process	Was the incident status section actively updated during the incident?	yes
	Was the public status page updated?	no
	Is there a phabricator task for the incident?	yes
	Are the documented action items assigned?	yes
	Is this incident sufficiently different from earlier incidents so as not to be a repeat occurrence?	yes
Tooling	To the best of your knowledge was the open task queue free of any tasks that would have prevented this incident? Answer “no” if there are open tasks that would prevent this incident or make mitigation easier if implemented.	yes
	Were the people responding able to communicate effectively during the incident with the existing tooling?	yes
	Did existing monitoring notify the initial responders?	no
	Were all engineering tools required available and in service?	yes
	Was there a runbook for all known issues present?	no
Total score (count of all “yes” answers above)		10