balance stuck at 0%, what can I do?

[Dewdman42]

I recently enabled scheduled Balancing on my readynas OS6.  It started the balance job last night at midnight as specified.  However, today I can't login via SSH and can't mount the SMB shares and all I can do is get to the frontview screen, which all seems to work, but it says currently balancing at 0% and basically everything such as timemachine has stopped working..aside from frontview..

what can I do to get this working?  I already tried to restart it from frontview, which I'm not sure it even did or not.

[StephenB]

Generally speaking, the longer a balance takes to run, the more it is needed. 

What firmware are you running?

You can check the status with ssh using

# btrfs balance status -v /data

(or substituting your volume name if that is different).

You can also manually cancel it using

# btrfs balance cancel /data

Another option is to pause it during the day, and resume it at night (letting it finish off-hours)

# btrfs balance pause /data
# btrfs balance resume /data

I'd also suggest taking a look at disk health, as that could explain the very slow speed.

# smartctl -x /dev/sda
# smartctl -x /dev/sdb

Adding the -x gives you information on disk errors that often doesn't show up in the SMART stats.

[Dewdman42]

I can’t do any of those things because ssh won’t work for me now. I get the “last logged in” message then it hangs there and never gives me a prompt.

Front view seems to work and says balancing is 0% done.

OS6, latest one

[StephenB]

Frontview runs at a lower priority than ssh, so it's odd that it is responding when ssh isn't.

Can you get the logs with RAIDar?

One option is to try a reboot, and see if you can get in with ssh quickly.

---

@Dewdman42 wrote:

OS6, latest one

---

Meaning 6.10.0

[Dewdman42]

Will try those things today. The whole device is mostly being used only as a time machine and smb backup device. I might just do a hard factory reset and make sure to set up scheduled balancing from the beginning.

It has been using a jbod raid, is there any reason to suspect that would cause a problem with balancing?

[StephenB]

---

@Dewdman42 wrote:

It has been using a jbod raid, is there any reason to suspect that would cause a problem with balancing?

---

No, not that I can think of.

I was wondering about a disk issue though, so if you do decide to go with a reset you might want to test them.

[Dewdman42]

Well the good news is that over night it completed...  all works fine now.  just took a long time.  no idea why it needed to turn off ssh and SMB while it was running, but maybe now its on a weekly schedule it won't take so long next time

[StephenB]

---

@Dewdman42 wrote:

 but maybe now its on a weekly schedule it won't take so long next time

---

I run it every three months (my schedule for all the maintenance tests).

It ran a couple weeks ago on my RN526x (18 TB volume, about 14 TB of data), and took about 20 minutes.

[Sandshark]

When one of my eSATA expansion chasses (one EDA500 and one other brand) do a balance or scrub, I have seen the readynasd task go to 100% CPU use in TOP, and then HTTP and SMB access stops.  I can only see it via SSH because i already have the session open when it happens.  And that's on an RN516, so I can imagine what would happen on the lower-powered Ultra.  When all is going right, the various kworker and such processes don't even take half the CPU throughput. 

When it doe shappen. I can typically pause a balance, let readynasd "calm down:", and then re-start the balance.  But there is no pausing a scrub.   I suspect that the readynasd process gets some kind of time-out in accessing the volume and then goes into a state it should not be in.  I figured it was because of the eSATA port expander in the external chasses creating a bottleneck, but maybe it can happen in other cases as well.

You may want to start an SSH sesson running TOP and then manually do a scrub to see if something similar is happening to you.