aboutsummaryrefslogtreecommitdiff
path: root/cmd/crawl/crawl.go
diff options
context:
space:
mode:
Diffstat (limited to 'cmd/crawl/crawl.go')
-rw-r--r--cmd/crawl/crawl.go2
1 files changed, 1 insertions, 1 deletions
diff --git a/cmd/crawl/crawl.go b/cmd/crawl/crawl.go
index ea88412..8c20901 100644
--- a/cmd/crawl/crawl.go
+++ b/cmd/crawl/crawl.go
@@ -37,7 +37,7 @@ var (
validSchemes = flag.String("schemes", "http,https", "comma-separated list of allowed protocols")
excludeRelated = flag.Bool("exclude-related", false, "do not include related resources (css, images, etc) if their URL is not in scope")
resumeDir = flag.String("resume", "", "path to directory of previous crawl to resume")
- warcFileSizeMB = flag.Int("output-max-size", 100, "maximum output WARC file size (in MB) when using patterns")
+ warcFileSizeMB = flag.Int("output-max-size", 5000, "maximum output WARC file size (in MB)")
cpuprofile = flag.String("cpuprofile", "", "create cpu profile")
bindIP = flag.String("bind", "", "IP address from which to make outbound connections")