GoReplay Logo
Operator docs for capture and replay. Product pages: PRO.

Save and replay from files

Job: record production HTTP to disk, then replay it later without touching prod again.

sudo gor --input-raw :8080 --output-file requests.gor
gor --input-file requests.gor --output-http="http://staging.example.com"

Replay preserves original gaps between requests unless you apply a percent limiter on --input-file (that speeds up or slows down instead of dropping).

Chunks and append

Default: each flush goes to a new indexed path. --output-file-append appends into one file per name.

# 20140608_0.log, 20140608_1.log, ...
gor --input-raw :80 --output-file %Y%m%d.log

# single file per day
gor --input-raw :80 --output-file %Y%m%d.log --output-file-append

Date tokens: %Y %m %d %H %M %S. Default name pattern when using those tokens is hourly (%Y%m%d%H) if you rely on the built-in helpers.

  • --output-file-size-limit — chunk size (default 32mb; k/m/g suffixes).
  • --output-file-queue-limit — chunk queue length (default 256).
  • --output-file-max-size-limit — max output file size (default 1TB).
  • --output-file-flush-interval — force flush (default 1s).
  • --output-file-buffer — temp dir for the current buffer (default /tmp). Used heavily for S3 in PRO.

.gz in the path enables gzip on write and read.

sudo gor --input-raw :80 --output-file /var/log/gor/%Y-%m-%d.gz --output-file-size-limit 256m --output-file-queue-limit 0

Reading

--input-file accepts a glob; files are replayed in lexicographic order. --input-file-loop restarts when the set finishes. --input-file-dry-run reports expected duration without sending. --input-file-read-depth (default 100) is the in-memory reorder buffer. --input-file-max-wait caps idle time between records. --input-file-exit-delay waits after the last record so responses can finish.

gor --input-file 'logs-2016-05-*' --output-http="http://staging.example.com"
gor --input-file requests.gor --input-file-loop --output-http="http://staging.example.com"
gor --input-file requests.gor --input-file-dry-run --output-null

File format

Plain HTTP plus a meta line (payload type, request id, timestamp). Records are separated by a unique banner line. Payload types: 1 request, 2 origin response, 3 replayed response. You can grep it; do not let an editor change newlines.

Faster than realtime

gor --input-file "requests.gor|200%" --output-http="http://staging.example.com"

More: load testing. S3 paths are PRO: PRO.

Replay options: replaying.