HDD at 100% when using rclone to copy to Backblaze

Help and Support
1

Hello everyone.

I have a server and I’m doing backup every 3 days to Backblaze. However, in the last month I start receiving notifications that my websites were going down when the rclone was working. Doing a research I found that the hdd is going over 100% busy . Is there are way to decrease this even if the upload speed will be reduced?

Right now I tried this:

/usr/bin/rclone --transfers=1 --buffer-size 5M --bwlimit 8M copyto /backup/ Backblaze:servers-backup/data

But the issue persists.

Thank you.

Kind regards.

2

Try reducing the number of --checkers to 1 and see if that helps.

3

I just tried…it’s still happening :frowning:
Thank you.

4

I think that’s normal as it’s checksumming the files. You can use ionice or something like that to make it no run at high priority.

I use this for my torrents.

 /usr/bin/ionice -c2 -n5 /usr/bin/rtorrent
5

You should also check your disk itself. If you have disk errors, they’ll manifest like that.

6

Thanks everyone. I tried again with both, nice and ionice and it’s still going to 101%. The HDD smart reporting everything is ok, but when I run badblocks the server goes down again, so I cannot finish the test without killing all the websites. What’s interesting, I’m coying right now manually through FTP the backups to my pc and even if I’m using about 11 MB/s bandwidth, the HDD is at maximum 35%. I don’t understand why badblocks or rcopy makes it going up to 101% with only 8 MB/s bandwidth.

//Later edit: local copy through FTP gone up to 102%…same issue.

7

Hi everyone,

I just tried to backup with qBackup and the hdd is going few seconds to 100% but most of time is at 60%.
However, that script doesn’t work for me since I want something to delete old files, not only to upload.

Last night I passed a badblocks test and the disk is okay.

Any idea what can I try with rclone?

Thank you.

8

Can you use ionice like I suggested?

9

I already tried and I still have the same issue. I tried both, ionice and renice and the hdd I/O still goes to 100%

10

ionice makes it so other things can use IO and if not, it can use up to 100%.

11

Already tried, even with n7 :tired_face:

101

12

Think of it like if you have 100% and if something else is requesting it, it gets priority, but the process can still use up to 100% so the screenshot looks right to me.

13

I’m gonna test again later when I will have few traffic on the websites. I’m scared are gonna give Error 500 again if I’m trying now.
I’ll be back with an update soon.

Thank you so far.

14

Hello @Animosity022,

I just tested it again and well…Same issue: Error 524 - A timeout occurred.

This is extremely frustrating, since I never had this issue before. I don’t know what to test

15

I was digging around a bit but I’m not sure offhand as on my system, ionice works properly although I’m on Debian.

When you run it and check the process, does it show the right #?

felix@gemini:~$ ps -ef | grep torrent
vpn      18046     1  0 Dec05 ?        00:00:07 /usr/bin/SCREEN -d -m -fa -S rtorrent /usr/bin/ionice -c2 -n5 /usr/bin/rtorrent
vpn      18047 18046  0 Dec05 pts/2    01:25:52 /usr/bin/rtorrent
felix    26349 26330  0 14:04 pts/0    00:00:00 grep torrent
felix@gemini:~$ ionice -p 18047
best-effort: prio 5
16

Yes, it’s showing everything fine

root@beta:~# /usr/bin/ionice -c2 -n7 /usr/bin/rclone --transfers=1 --checkers=1 --buffer-size 10M --bwlimit 6M copyto /backup/ Backblaze:servers-backup/beta

root@beta:~# ps -ef | grep rclone
    root     16857 16796 45 20:16 pts/3    00:00:04 /usr/bin/rclone --transfers=1 --checkers=1 --buffer-size 10M --bwlimit 6M copyto /backup/ Backblaze:servers-backup/beta
    root@beta:~# ionice -p 16857
    best-effort: prio 7

Also after I’m stopping the rclone it takes like 5 minutes to recover and to get hdd I/O below 90%

17

I ran it a few times using the stress command and I can peg the disk IO, which is fine and it lets other processes take the disk IO if they come on.

I did it by using:

root@gemini:/data stress -d 5
stress: info: [31087] dispatching hogs: 0 cpu, 0 io, 0 vm, 5 hdd
^C
root@gemini:/data /usr/bin/ionice -c2 -n5 stress -d 5
stress: info: [31140] dispatching hogs: 0 cpu, 0 io, 0 vm, 5 hdd
^C
root@gemini:/data

I checked using iotop and once the process was ionice’d, I did a cp on the same drive and that took priority.

18

I tried with stress and ionice and the hdd crash the server… :frowning:
It reached 103% and 140 MBw/s

I tried everything my friend.

gtmetri

19

I don’t think you have a rclone issue per se as if you can recreate it using stress and it still breaks, something else has to be going on.

stress even at 100% shouldn’t crash the server but should make it slow.