What kind of Plex performance should I expect with gdrive rclone+mergerfs?


#1

So I’m using rclone with a gdrive Team Drive, behind a mergerfs.

What kind of performance should I expect?

For example: Sometimes, an episide of a show will take 1-2 minutes to start playing, but then it will play smoothly without any problems after that.

Mergerfs script:

mergerfs \
-o defaults,sync_read,allow_other,category.action=all,category.create=ff,func.getattr=newest \
/home/user/media:/home/user/gdrive \
/home/user/mergedmedia

Rclone mount:

rclone mount \
--allow-non-empty \
--allow-other \
--bind <ip here> \
--buffer-size 256M \
--cache-tmp-wait-time 5m \
--dir-cache-time 72h \
--drive-chunk-size 32M \
--fast-list \
--log-level INFO \
--log-file /home/user/reboot_scripts/rclone.log \
--poll-interval 5m \
--size-only \
--timeout 1h \
--tpslimit 10 \
--umask 002 \
--vfs-read-chunk-size 128M \
--vfs-read-chunk-size-limit off \
gdrive: /home/user/gdrive

Is it normal that some items will take that long to start playing? This is original quality, no transcoding.

All info appreciated! <3


#2

Its not normal, in most cases the video should start within 5 to 10 seconds.
Its more then likely you do not use your own API key with gdrive. ( i suggest you remake rclone config and add those keys )

You could also do some seektesting with this script


Just modify file/paths

My results:
SEEKSPEED /mnt/gdrivecrypt/tmp/1G.file1
2019/02/11 07:46:42 Reading 527 from 134020434
2019/02/11 07:46:44 Reading 187 from 647328285
2019/02/11 07:46:44 Reading 390 from 164029137
2019/02/11 07:46:45 Reading 684 from 957743134
2019/02/11 07:46:46 Reading 932 from 958990240
2019/02/11 07:46:47 Reading 687 from 864509188
2019/02/11 07:46:48 Reading 1009 from 978417974
2019/02/11 07:46:49 Reading 282 from 1018408203
2019/02/11 07:46:49 Reading 149 from 1011747688
2019/02/11 07:46:50 Reading 994 from 627253638
2019/02/11 07:46:51 Reading 218 from 379326753
2019/02/11 07:46:52 Reading 658 from 915302975
2019/02/11 07:46:53 Reading 299 from 752835466
2019/02/11 07:46:53 Reading 248 from 97964648
2019/02/11 07:46:54 Reading 1015 from 913623924
2019/02/11 07:46:54 Reading 888 from 999577560
2019/02/11 07:46:55 Reading 783 from 960629247
2019/02/11 07:46:55 Reading 332 from 438689995
2019/02/11 07:46:56 Reading 1015 from 919928731
2019/02/11 07:46:57 Reading 914 from 449922548
2019/02/11 07:46:58 Reading 658 from 1064433867
2019/02/11 07:46:58 Reading 835 from 1019929124
2019/02/11 07:46:59 Reading 915 from 1033801660
2019/02/11 07:46:59 Reading 740 from 909794159
2019/02/11 07:47:00 Reading 89 from 1015643398
2019/02/11 07:47:00 That took 17.871803059s for 25 iterations, 714.872122ms per iteration
Finished in 0m:19s

This shows how fast was seek processing file in 25 iterations, check the number per iteration and compare how log did it take vs Finished, as you can see it took cca 1.2 seconds for API to reply ( you will probably have here larger delay )

p.s. From your mount script you can drop fast-list (does not do anything with mount), not sure why you need size-only (good for copy/sync)


#3

Thank you so much for your help. I will look into whether I’ve set up the API key properly and report back.


#4

Yup that was it. This is so much more insanely faster.