I've been getting complaints from many of my Plex users about stuttering, and I've been having the issue myself for quite some time. It's not just noticeable in Plex, when I´m browsing shares on the unRAID server I get random slowness and timeouts, and some servers even lose connection with the mapped network shares which causes all sorts of problems with various software.

 

I've got 3 physical servers, unRAID being one of them, and multiple VM's on the other servers and I've been pinging from server to server and all servers respond in the >1ms range, as expected, all server connected to the same Gbit switch. However, this is a small example of the results from the unRAID server. The only hardware aspect I could change was the network cable and switch port, that changed nothing. The stuttering in Plex is in check with the highest spikes, or when the spikes last for 10s or longer and the buffer isn't enough.

 

Reply from 192.168.1.30: bytes=32 time=1259ms TTL=64
Reply from 192.168.1.30: bytes=32 time=1124ms TTL=64
Reply from 192.168.1.30: bytes=32 time=1792ms TTL=64
Reply from 192.168.1.30: bytes=32 time=1760ms TTL=64
Reply from 192.168.1.30: bytes=32 time=1288ms TTL=64
Reply from 192.168.1.30: bytes=32 time=1903ms TTL=64
Reply from 192.168.1.30: bytes=32 time=1518ms TTL=64
Reply from 192.168.1.30: bytes=32 time=1282ms TTL=64
Reply from 192.168.1.30: bytes=32 time=1109ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time=1483ms TTL=64
Reply from 192.168.1.30: bytes=32 time=1534ms TTL=64
Reply from 192.168.1.30: bytes=32 time=1674ms TTL=64
Reply from 192.168.1.30: bytes=32 time=1408ms TTL=64
Reply from 192.168.1.30: bytes=32 time=1010ms TTL=64
Reply from 192.168.1.30: bytes=32 time=773ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time<1ms TTL=64
Reply from 192.168.1.30: bytes=32 time=998ms TTL=64
Reply from 192.168.1.30: bytes=32 time=1573ms TTL=64

 

 

This goes on and on, even when there's no traffic at all from the server. Goes to <1ms for a couple of seconds, then spikes.

 

What in the world could be causing this? This is just about making my unRAID server unusable.

What networking hardware do you have between your unraid server and your test points? What make/model of network switches or routers? There have been some troublesome switches that can exhibit these symptoms due to dropped packets or mishandling of them.

Cheap noname dumb planet 16port gbit switch.

Would be curious if this problem occurs for you with unRAID 6.  Not suggesting you have to upgrade to fix it, but many users have reported performance increases in making the move.  UnRAID 6 features a few key changes including the move to a 64bit OS and support to run applications like Plex as Docker Containers instead of a tar.xz package for Slackware.

I've been looking into upgrading, just don't feel the need to just yet. I don't run anything else on the box but unRAID and that's not something that will change, even if unRAID can do virtual now. I like to keep my fileserver just doing that, serving files, and that's it.

 

I will def. upgrade at some point but since the server is heavily used pretty much 24/7 it's not something I´m going to rush into. This issue is relatively new, so it's very unlikely that it's a version specific issue.

 

I've exp. some slowness in the past when the mover is working, and oddly I find the whole system run a bit slower when drives begin to fill up, but there's nothing in the system log that gives any indication as to what could be causing this. What I´m pretty much asking is, what would be the next logical step to do?

 

 

I've been looking into upgrading, just don't feel the need to just yet. I don't run anything else on the box but unRAID and that's not something that will change, even if unRAID can do virtual now. I like to keep my fileserver just doing that, serving files, and that's it.

 

We will be introducing a unRAID 6 "lite" mode that will address this for people like you:  all the file serving feature goodness you want, none of the virtualization required.  That said, you should know that the primary reason for someone like you to upgrade is better performance and better support.  UnRAID 6 is based on an updated and  64bit base OS and many users have reported an overall performance increase compared to previous versions.

 

All of that said, are you running Plex on another machine then and letting it connect to unRAID over the network?  I guess I'm a little confused because you say you don't want your NAS doing anything else, but you are mentioning an issue with Plex.  If you are running Plex on unRAID, you might as well be virtualizing with either docker or virtual machines.  Virtualizing Plex is safer for your server given that you can isolate Plex's access to vital system resources like memory.  Native application installs through Plugins cannot boast this claim.

Sorry if I wasn't clear enough

 

The unRAID fileserver doesn't run anything but a plugin or two for monitoring and alerts. I also forgot to mention that I´m running the Dynamix WebGUI, if that can have anything to do with this.

 

The virtual servers run everything in VM's - Plex, AD, DNS, VPN, FTP, p2p software etc.

 

I like the sound of the v6 lite version, I´ll probably wait for the release of it before upgrading. I just tried putting in a new NIC, just to see if it could possibly be a funky hardware issue but it remains the same.

Sorry if I wasn't clear enough

 

The unRAID fileserver doesn't run anything but a plugin or two for monitoring and alerts. I also forgot to mention that I´m running the Dynamix WebGUI, if that can have anything to do with this.

 

The virtual servers run everything in VM's - Plex, AD, DNS, VPN, FTP, p2p software etc.

 

I like the sound of the v6 lite version, I´ll probably wait for the release of it before upgrading. I just tried putting in a new NIC, just to see if it could possibly be a funky hardware issue but it remains the same.

Gotcha, OK, so you're kinda making this harder on yourself.  The media content needs to stream from your NAS, to your VM host, then back out from the host to your playback device.  With Plex on unRAID and in a container, you completely eliminate one hop through the network in this equation.

 

That said, if your NAS is not sufficiently stocked with a strong enough processor or memory (e.g. a low end / old hardware), then it might not be powerful enough to handle transcoding content.  That said, I've had good success with transcoding on fairly low end systems.

 

As far as Dynamix or any other plugins affecting you, the simple test to validate would be to boot into unRAID safe mode and attempt your test again.  This disables all plugins temporarily and is a good way to narrow down the root cause of an issue.

 

Also, your cheap switch could be part of the problem too...

I´m aware of the extra length the stream has to go. Having a standalone fileserver is just something of a good practice of mine. Keeping things seperate makes it easier to diagnose problems, revert the VM's back if serious problems occur, having a failover option when clustered etc. It's also a matter of hardware, my current unRAID machine is an older 775/C2Q/8GB RAM box, plenty for unRAID but not exactly much of a VM server.

 

I've got about 50+ users accessing my content, most of them through Plex but some over FTP or VPN in rare occasions, so I need to keep my Plex setup on the 8 core Hyper-V machine. It probably adds to the equation that I´m an MCITP, so I´m much more comfortable troubleshooting things in the M$ enviroment than *nix based systems, and with that userbase I don't like to be at the mercy of google/forums

 

I don't really have much faith in those consumer switches but.. unRAID is the only machine giving me those random spikes, every other machine in the house that's Gbit wired (and there are about 7 of them) all respond within the <1ms range.

 

I completely forgot about the safe mode option! I'm going to try that early in the morning when there's little to no traffic, thanks for the suggestion.