Posts by grokov

IIRC OMV update is the same as system update with apt. I'm guessing legacy refers to kernel 4.

If it works, it's not a wrong version. I'm not sure if anything about OMV5 requires armbian buster / 5.4.y. As far as I can tell, OMV and armbian are independent-ish, with OMV being packages and configurations applied on top of an OS like armbian or others. Kernel 5.4.y is the latest with armbian and makes sense to use it in the longer term. armbian-config lets you choose different kernels, but I have no idea how well that would work in your case.

Yes, I saw and tried applying it too through armbian-config. You can ask on IRC #armbian, but last time I checked, they said they didn't have time yet to work on kernel 5.4.y configs for the HC2. In fact they said they're not happy with the state of the performance and know it can be improved, but again it's a question of time/effort. At least it's stable.

No, to my knowledge the board optimizing configurations have not been implemented in the latest armbian, so applying them is pointless.

This issue was with a big performance regression in memory copy speed, which has thankfully been fixed. Just make sure you're on a recent kernel, and it's all good now.

If you're confident the issue is actually with folder2ram, it would be good to open an issue for it: https://github.com/OpenMediaVa…avault-flashmemory/issues

Quote from Offal

When I reboot, folder2ram is still not syncing ram to the filesystem. I would have thought that -umountall will sync before it umounted.

Running 5.5.5-1 (Usul) with folder2ram.

How can you tell it's not syncing?

With OMV 5.5.5-1 and openmediavault-flashmemory 5.0.7 (folder2ram plugin), it has been ok-ish for me. In some cases data is lost on reboot, but most times it persists. I haven't found a pattern.

iperf and iperf3 are not intercompatible. Use iperf3 -s

https://serverfault.com/questi…failed-connection-refused

Fortunately a workaround has been found. Currently on kernel 5.4.50-odroidxu4 the performance has been restored. http://ix.io/2qJf

Note that there is a serious performance regression with HC2/HC1/XU4 with kernel 5.* where the memory bandwidth and latency are multiple times worse than on kernel 4.*

See https://forum.odroid.com/viewtopic.php?f=184&t=38774

You can use the benchmarking tool sbc-bench (also available through sudo armbian-config -> Software -> Benchmarking) and post your results. Kernel 4.* results in particular would be interesting.

The performance data has persisted over the past several reboots over the past month or so. I don't know why, whether it's something I did or a bug was fixed in updates, but at least it seems ok for now.

I have the WebGUI Auto logout timeout set to 60 minutes, and previously this held, but lately it is timing out after 5-10 minutes.

More precisely, the WebGUI appears to be logged in, but attempts at navigating fail, and a few seconds later, the flashing red and black screen shows up requiring a left click to go back to the login screen.

I changed and applied the settings to 10 minutes, and then again back to 60 minutes in case it had not been properly set or applied. No change. I also confirmed that the setting is correctly reflected in config.xml.

Is anyone else having this issue?

So bizarre and frustrating. I rebooted again, and this time the data since the last reboot is gone. But some previous data prior to the last reboot is still there. Very confusing!

Hmm... I just tried rebooting after about a day of having tried your fix, and so far the data survived. Btw, my /var/lib/collectd/rrd/* is still empty due to the previous fix I tried and I have not (yet) reversed that change.

Did you do systemctl daemon-reload after making the changes to /lib/systemd/system/folder2ram_shutdown.service?

I've forgotten the behavior in OMV4+Armbian, but in OMV5+Armbian, previous boots don't have entries in journalctl:

$journalctl -b -1
Specifying boot ID or boot offset has no effect, no persistent journal was found.

Apparently persistent journal is not necessarily enabled in /etc/systemd/journald.conf and I confirmed that /var/log/journal does not exist. https://askubuntu.com/question…ter-ubuntu-16-04-restarts

Is this intentional?

Thanks, I'll try ExecStartPre=/sbin/folder2ram -syncall.

I already tried the first solution of deleting/moving the /var/lib/collectd/rrd/localhost directory which contains .rrd files (much like in /var/lib/rrdcached/db/localhost) as mentioned at RE: error in syslog regarding rrdcached which I found since I had the same type of system log errors mentioned in RE: error in syslog regarding rrdcached. After this, the data seemed to survive a couple of test reboots and I thought it was solved, but after a day, rebooting again wiped out previous data.

BigBro Does your problem go away when you clear /var?

Mine does not look full:

Filesystem      Size  Used Avail Use% Mounted on
udev            927M     0  927M   0% /dev
tmpfs           200M   26M  175M  13% /run
/dev/mmcblk1p1   59G  1.6G   57G   3% /
tmpfs           997M     0  997M   0% /dev/shm
tmpfs           5.0M     0  5.0M   0% /run/lock
tmpfs           997M     0  997M   0% /sys/fs/cgroup
tmpfs           997M  4.0K  997M   1% /tmp
folder2ram      997M  129M  869M  13% /var/log
folder2ram      997M     0  997M   0% /var/tmp
folder2ram      997M  888K  996M   1% /var/lib/openmediavault/rrd
folder2ram      997M  572K  997M   1% /var/spool
folder2ram      997M   35M  963M   4% /var/lib/rrdcached
folder2ram      997M  4.0K  997M   1% /var/lib/monit
folder2ram      997M  1.3M  996M   1% /var/cache/samba
/dev/sda        9.1T  807G  7.8T  10% /srv/dev-disk-by-label-WD100EFAX
tmpfs           200M     0  200M   0% /run/user/1000

Very frustrating. I still haven't found a solution to this.

Quote from grokov

Does anyone else have journalctl entries available only since the current boot? (on OMV5 installed on an SD card using Flash Plugin)

I made a post about Performance Statistics getting cleared on each reboot, but it's not just those stats but system journal entries too. Performance Statistics losing data after reboot

Quoting myself... since there is no reply about this problem, is the lack of persistence likely not due to OMV5?

Maybe the issue is with Flash Plugin? Or maybe it's something with rrdcached? Is there a proper way to reset rrdcached statistics data and journal?

If it's not related to OMV5 maybe I can ask on a different forum, if I can get a clue of what is likely responsible for the problem.

It turns out the Optimized Board Configuration for the HC1/HC2 is not prepared for kernel 5.4.y yet. So for now it is best to stick with the XU4 configuration for the full 2000 MHz and correspondingly improved performance:
http://ix.io/2k57

Use systemctl list-files --state=failed to see which services had issues. On my Odroid I have watchdog.service and wd_keepalive.service which fail with similar errors. I'm not sure if there are modules available for it.

To disable the watchdog in this case: sudo systemctl disable watchdog.service and sudo systemctl disable wd_keepalive.service

Does anyone else have journalctl entries available only since the current boot? (on OMV5 installed on an SD card using Flash Plugin)

I made a post about Performance Statistics getting cleared on each reboot, but it's not just those stats but system journal entries too. Performance Statistics losing data after reboot