Field notes · Troubleshooting
Every post starts as a support ticket or an on-premise install: the symptom I saw, the metric that proved it, and the fix that shipped to the customer.
A Java mail client ignored mail.smtp.auth=false and failed with 'AUTH LOGIN failed' on a relay that required no authentication. Reproducing it with MailDev, and why blank credentials are not the same as no credentials.
· 6 min read
No guides match that filter. Clear the text or pick another tag.
Coverage
Tags come straight from each post’s frontmatter, so this table is the index itself, re-counted. Bar length is relative to the busiest tag.
How I write
Not "the service was slow" but "p99 rose from 240 ms to 4.2 s between 02:10 and 02:40". The threshold that triggered the alert is part of the symptom.
Each claim points to the chart, trace or log line I used, so you can reproduce the evidence in your own project instead of trusting mine.
I show the same chart after the change. If the fix only moved the problem, the guide says so and stays open.
A short "prevent this again" section with the alert, dashboard widget, or URL check that would have caught it earlier.
This index · personal field notes
troubleshooting post published, each one rendered from a markdown file in my notes vault.
distinct tags across the index, taken verbatim from each post’s frontmatter.
median read time, computed from each post’s prose word count at 200 words per minute.