Failed to drain template buffer and horrible rclone serve performance

Using rclone serve http, with a google drive backend on a 10 gbps server, and connecting to them with 1 gbps servers in the same datacenter, I have horrible performance ( 14 MBytes/s ) and I also see this error of the title in the logs.

Also I see lots of files in the logs forever stuck at "transferring"

Any ideas why? I'm doing this to save on the API hits, the idea would be that the 10 gbps server pulls from google, and then serve to others servers behind it.

If I use google drive as the backend on the 1 gbps servers, everything works as usual.

My settings on the mount:

  --user-agent='Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/74.0.3729.131 Safari/537.36' \
  --allow-other \
  --fast-list \
  --vfs-read-chunk-size=10M \
  --vfs-read-chunk-size-limit=0 \
  --buffer-size=0 \
  --poll-interval=1m \
  --no-modtime \
  --drive-pacer-min-sleep=10ms \
  --dir-cache-time=24h \
  --timeout=10m \
  --umask=002 \
  --union-search-policy=eprand \
  --log-level DEBUG \
  --log-file /opt/rclone.log \
  --async-read=false \
  --rc \
  --rc-addr=localhost:5572 \

on the rclone serve settings:

/usr/bin/rclone serve http union: \
  --addr :8080 \
  --no-modtime \
  --vfs-read-chunk-size 10M \
  --vfs-read-chunk-size-limit 0 \
  --timeout=11m \
  --contimeout 3m \
  --log-level INFO \
  --log-file /opt/24hs.log 

I was expecting this to have really great performance, as I can download from google on the 10 gbps server at almost 100% speed and also have practically local network connection to the other servers connecting to the rclone serve.

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.