NAS 104 locking up

Hi!

I think I've fixed my RN104-box by changing the swappiness-value. Per (readynas-)default the swappiness is set to "0". This means that swapping will only be done if there is no more RAM left. This is - in my oppinion - a bad decision. Because of this slow onboard-controller there maybe is not enough time to swap-out when your processes hit the end of the RAM. My MySQL-instance always got kicked which shot the rsyslog-daemon to death and so on...

I changed the swappiness by adding a row to the end of the file /etc/sysctl.conf:

swappiness = 60

100 = swap everything (= also stupid and makes everything very slow) 60 = Linux default 0 = no swap - only swap if there is no more RAM (readynas-default)

Changing this value makes the setting reboot-aware / persistent.

You have to be "ROOT" (via ssh) to change these settings.

You can use this statement on the commandline :

echo swappiness = 60 >> /etc/sysctl.conf

This change is static and has only to be done once...

You'll have to reboot after changing this settings.
__________________________________________________

If you are not brave enough to change this system-file or want to check the behavior you can use following statement:

sysctl -w vm.swappiness=60

Remember: This is a dynamic change and has to be done every time after a reboot!!!

Have fun and good luck,
Roy

Agreed on the swappiness: viewtopic.php?f=18&t=74045

Another BIG problem is this (sorry) stupid minidlnad that seems not to be designed for single-core-cpus. Disabling ReadyDLNA (=minidlnad) gives you a HUGE performance-boost.

You can use this little commandset to check the "death/sleeping" processes that are causing the IO-problems on this box:

# for HH in $(seq 1 1 10); do ps -eo state,pid,cmd | grep "^D"; echo "---========================================---"; sleep 5; done

For example:

root@readynas1:/tmp# for HH in $(seq 1 1 10); do ps -eo state,pid,cmd | grep "^D"; echo "---========================================---"; sleep 5; done D 7672 chattr -c /data/sabnzbd ---========================================--- D 1355 [btrfs-transacti] D 7684 /usr/sbin/minidlnad -S ---========================================--- D 7689 rm -rf /apps/.readydlna/files.db /apps/.readydlna/art_cache ---========================================--- D 1922 [flush-btrfs-1] D 7695 /usr/sbin/minidlnad -S ---========================================--- D 772 [jbd2/md0-8] D 1922 [flush-btrfs-1] D 7695 /usr/sbin/minidlnad -S ---========================================--- D 1922 [flush-btrfs-1] D 7695 /usr/sbin/minidlnad -S ---========================================--- D 1922 [flush-btrfs-1] D 7695 /usr/sbin/minidlnad -S ---========================================--- D 7695 /usr/sbin/minidlnad -S ---========================================--- D 1355 [btrfs-transacti] D 7695 /usr/sbin/minidlnad -S ---========================================---

To make miniDLNA sweat copy some JPEGS or movies onto a DLNA-enabled share while minidlna is running and watch the IO-wait grow into heaven.
You also can disable DLNA while copying MANY files onto the box, what in my opinion makes no sense, but maybe helps.

Kind regards,
Roy

Good thing I found this thread because I was about to return mine! I've only had it for about 5 days and it was being insanely slow when trying to transfer movies and tv shows from my pc to the NAS, it didn't matter if it was via NFS, SMB, FTP or RSYNC. It has crashed on me about 6 times already and as a previous poster stated the only way to revive it is to pull the plug. It just kept on saying that it lost the connection and I thought my router was the problem since I constantly have problems with it but it is most likely the DLNA daemon (along with the swappiness setting, but my RAM is usually around 350 MB used) since I was/am copying terabytes of data. I got pissed off and deleted most of it because after constantly transferring for two days it wasn't even done so I just said "F*** it!" and wiped the drive and put it in the NAS, which automatically set it to RAID1 which kind of annoyed me (would rather it be in RAID0 until I get my two other drives and can put it in RAID 5 or 10). I'm downloading stuff with SABnzbd and the CPU is always pegged at 100% and the temperature is around 150F. I'm at work now and can't test this out because surprise surprise the damn thing crashed again and I can't SSH or use the ReadyNAS remote to get in.

edit: this didn't help. I changed swappiness to 60 and DLNA isn't enabled. I left it on over night to copy files over and then had SAB auto-resume after it was done copying the files. I woke up 6 hours after I had started this to find that everything was still working (but really slowly) and that it had actually swapped out data from the RAM...but then it proceeded to crash constantly after I started interfacing with it.

Same problem here - have a support ticket open ( #22378897 ) they tried to close it as 'unsupported drives' first because I got the 4T version of the 3T Hitachi drives - replaced them with 5900RPM Seagates that are on the supported list.

Been insulted continually by the support person with regards to my gender.

He just told me disable bonding because of a dropped packet (one)..

Already worked out the problem is some process running wild ('top' will show you an 'every second' CPU usage - and hitting 40.7 at one point was one give away) ... Noted that when locked up the last status of running processes in top showed massive issues with IO.. but no process showing as high usage.

Disabled the Anti-Virus (real time AV) scanning and went from lockups every 2 hours to every 10-14 hours.

This morning finally worked out how to disable the DNLA server and it's been running stable since.

Also noted that no swap is ever used (now I know why - setting swappiness soon) - also noted in logs that the OOM killer is nuking just about everything when the issues occur...

Netgear support has no clue so far.. I'm staying quiet in the hope that they might get to it themselves and as such they might actually fix it (though I am not holding my breath...)

Michelle

Wow that's pretty horrible. I don't have AV scanning enabled and as far as I know I don't have the DLNA server enabled because the little bar under DLNA isn't green where all the sharing protocols are listed. Also ps ax|grep -i dlna doesn't give me anything so I think it's safe to assume that it isn't running.

What log specifically shows what OOM is killing?

Here's the link to my thread: https://www.readynas.com/forum/viewtopic.php?f=24&t=74371&p=413290#p413290

Same problem here.

I (eventually) got to chat with a real support person yesterday* and he determined from the logs that the problem was with CPU.

He advised disabling snapshots on the shares (I have done).

I have also turned off DLNA as per recommendations above. Also AntiVirus is disabled (always has been), in fact the unit is running the bare minimum of services required, to minimise load.

The CPU usage has dropped from the high 90%s to the mid 80%s, still ridiculously high for a NAS with a mere 100GB of office-type documents used by half a dozen people, and spread across 6TB of storage. It's not streaming any media, running games, recording video, running databases, no apps installed.

It crashed again at 02:19, after 17h20 of uptime. Which is better, but not exactly world class.


In a nutshell, the support chat can be sumarised as:

"Your NAS is fitted with a CPU that can't run even half the things the NAS is sold to do. Please turn off some of the services you might reasonably expect a NAS to have available and that we advertise as features."

Utterly, utterly, unacceptable. This device appears unfit for purpose.

*After about an hour of reloading the 'ask for chat' page, waiting for it to say anything other than "We're too busy", then getting into the queue and spending another 15 minutes or so in said queue... Netgear, you need more support engineers!!

I'm returning mine since I bought another one from another company that has an 12. GHz Intel dual-core Atom processor and 2 GB RAM, it can also do HDMI out and runs XBMC! All for $100 more. I also left a bad review on New Egg saying that this thing is extremely underpowered and pretty much can't do anything.

Ditto - for not much more you can have the ReadyNAS 314, also with a 1.2GHz dual core Atom and 2GB RAM.

Shame, because the Ultra4, Ultra6, 516, and the 2100 and 2120 (rack mount) ReadyNAS units I've also used have all been really impressive.

Seems that Netgear have put together a "cheap" unit, that's just had too many corners cut.

:(

The next firmware 6.1.5 should have some improvements. It would be recommended to backup your data, update to 6.1.5 (once it is released), verify update is successful, do a factory default (wipes all data, settings, everything) and then restore data from backup.