Plex buffering issues

[Joseph Stacklin]

[Joseph Stacklin]

Looks like the cache does help after all

[Joseph Stacklin]

Update: When playing back locally (through RDP in IE) Plex will play (and buffer) the optimized version of Episode 1 of Star Trek Discovery (the exact same file I have been testing from the get-go). However when playing through the Plex app on my client machine (Windows 10 Pro) the system fails to play or buffer the video file. Video was transcoded at 720p 2.1Mbits/s using Plex.

 

Edited August 31, 2018 by wolfboytech
Incorrect bitrate

[Joseph Stacklin]

Update: The files play fine through Plex's web app (for longer than on the desktop one)

[Mikensan]

Just remember anytime you direct play, it is only using read. If you transcode, it reads > writes > reads which even for a single stream can be hard on a disk. When you say plex app, are you talking about plex media player? 

 

When you play through the browser, is it transcoding?

 

That said, is all of your network on the same subnet (client PC / server)? Everything connected at 1gbit?

[Joseph Stacklin]

3 minutes ago, Mikensan said:

Just remember anytime you direct play, it is only using read. If you transcode, it reads > writes > reads which even for a single stream can be hard on a disk. When you say plex app, are you talking about plex media player? 

Yes, the plex media player for windows which you download from their site, not the Plex Pass version from the Windows Store

3 minutes ago, Mikensan said:

When you play through the browser, is it transcoding?

No. all files were transcoded/encoded beforehand by plex

3 minutes ago, Mikensan said:

That said, is all of your network on the same subnet (client PC / server)? Everything connected at 1gbit?

Yes, everything is on 10.0.2.X/24 with the server having a theoretical output/input of 2Gbits/s with the clients having 1Gbit/s

[Mikensan]

2 minutes ago, wolfboytech said:

No. all files were transcoded/encoded beforehand by plex

You mean via the "optimize" feature? Depending on what you do you can still technically force it to transcode by selecting a non-native resolution (720p when the video is 1080p).

 

4 minutes ago, wolfboytech said:

Yes, everything is on 10.0.2.X/24 with the server having a theoretical output/input of 2Gbits/s with the clients having 1Gbit/s

Is this because you're using teaming/LACP on the plex host? Might not hurt to try disabling a card / teaming, as I'm not quite sure how plex would handle it - just for troubleshooting.

 

[Joseph Stacklin]

1 minute ago, Mikensan said:

You mean via the "optimize" feature? Depending on what you do you can still technically force it to transcode by selecting a non-native resolution (720p when the video is 1080p).

Yeah through the optimize feature. I selected the "play version" option when playing the videos

1 minute ago, Mikensan said:

Is this because you're using teaming/LACP on the plex host? Might not hurt to try disabling a card / teaming, as I'm not quite sure how plex would handle it - just for troubleshooting.

Yeah, I'm using a Windows Bond on the NIC's. I've tried it with and without (as mentioned in my first post) with no change in performance. I am getting some Mini-SAS to SATA cables and gonna try this again on my R410 (I just hate it because of the noise in comparison to the C1100). They're coming around the 5th so I will be able to test it then.

[Mikensan]

Just posting my results to compare, the write speeds of your C drive is what catches my attention. Oddly enough what supposed to be my "crappy" disk, the vertex, performed the best. And my USB cheapy external out performed my WD black :'(

 

So to summarize your issues of right now, direct playback from the file is fine. Direct playback over the web is fine. However playing over plex media player will not play?

 

Also for what it is worth, I also have a C1100 - I have 2x L5520s. I ran plex on this server for a while as a VM, assigned 16 CPUs to it though.

 

 

[dalekphalm]

15 hours ago, wolfboytech said:

Looking back at these results, they are realllllllllllly bad for an SSD.

 

Your Sequential Read/Write speeds should be 300+ MB/s - though 150 MB/s is certainly usable.

 

But your 4K random read and write speeds are actually frighteningly bad. They're only marginally better than a HDD. You should be getting 80 MB/s to over 300 MB/s here.

 

Now I did try and look up some reviews for the Sandisk P4 - it seems to be an SSD supplied to OEM's only. The only review for it I could find is for an Acer EeePC, and the results were pretty bad (look under EEE Slate PC):

https://www.tomshardware.com/reviews/eee-slate-ep121,2986-8.html

 

Do you have another drive to pop in and compare against? Another SSD or even a HDD?

[Joseph Stacklin]

Finally I have had the time to sit down and test the new server. The first result (left) is without caching enabled. The second (right) is with 8GB of DDR3 ECC RAM dedicated to PrimoCache. Plex is still having issues with or without the cache enabled so that rules that out. The 10TB array (with the plex libraries on it) for some reason is not able to be tested by Crystal DiskMark and crashes the entire system until I stop the test. The RAID 5 array is created on my PERC 7 card (512mb cache model) in Write-Through mode. I have 14GB RAM dedicated to the disk and disabling PrimoCache doesn't do anything with the speed of the system. I'm at a complete loss. Even with the system performing 100x its prior speed I am still unable to use Plex server for some reason, even on my local network.

 

I'll attempt a RAID 0 Array with just the data on it and see where we go from there!

 

[Joseph Stacklin]

I ran a max-sized byte test using ping running simultaneously on two machines (another server and my personal computer) and it doesn't seem to be a network issue as everything averages around 1ms with it maxing out at 9ms. So internal network is good. Just to make sure tho, I will connect directly to the router and bypass my dumb switch

 

[Joseph Stacklin]

Update: No.. Connecting directly to the router does absolutely nothing for difference. Takes just as long to load and still has stuttering and constant buffering. Investigating cables and backplane and will update after tests.

[AbsoluteFool]

On 27.8.2018 at 5:13 PM, wolfboytech said:

Hey all, so I have an up-to-date Plex server with Windows Server 2016 Datacenter Gen2 installed. It has 72GB of RAM (~50GB of which is being used as a cache system with PrimoCache), 2x Xeon E5620 CPU's, and a bridged NIC (I have tried with and without bridged NIC's and the bridged one has the least buffering time). The server is DMZ'ed from the outside and the ports are forwarded. Firewall is allowing Plex and all its services to send and receive data on the server and the client. This  issue happens across multiple clients and hasn't happened before on previous Plex servers on my network. Transcoding is set to "Prefer higher speed encoding". No GPU acceleration is being used due to no GPU being installed.

 

If this is an issue with Windows, I will gladly switch to a Linux distro but I just want to make sure that it isn't something I have done to mess with the configuration before I make the switch to the Linux distro and have the same issues. Any help is much appreciated.

Why would you have DMZ and port forwarding? (Just curious..)

As you say you have multiple clients hitting the server. Is this internal hits to the server or outside hits to the server?

The most common buffering problems with plex is the storage or cache drives. 

 

I wouldn't bet on it being Windows, i've had exactly the same problem on Linux. However it seemed that it was only a few selected movies that started this buffering issue on my side. Perhaps something you could check into?

[Joseph Stacklin]

10 minutes ago, AbsoluteFool said:

Why would you have DMZ and port forwarding? (Just curious..)

I have tried with PF and DMZ separate but Plex refuses to allow external connections for some reason. I also now only have it DMZ as I use it to pivot onto other devices on my Intranet from outside my house using RDP.

10 minutes ago, AbsoluteFool said:

As you say you have multiple clients hitting the server. Is this internal hits to the server or outside hits to the server?

The most common buffering problems with plex is the storage or cache drives. 

Outside the server on the Intranet.

10 minutes ago, AbsoluteFool said:

I wouldn't bet on it being Windows, i've had exactly the same problem on Linux. However it seemed that it was only a few selected movies that started this buffering issue on my side. Perhaps something you could check into?

Hmm.. Thanks for the info. Honestly I think it could be an issue with one of three things:

1. PERC Card (RIP the dream if that's the case)
2. One of the 5TB drives is failing (least likely due to them working in other machines, even in software RAID 5's)
3. Hardware inside the server (ie: backplane, SAS cables, etc.) <---- I find this one to be the most likely culprit since I can no longer create new arrays and my old array is very unstable.

[AbsoluteFool]

1 minute ago, wolfboytech said:

I have tried with PF and DMZ separate but Plex refuses to allow external connections for some reason. I also now only have it DMZ as I use it to pivot onto other devices on my Intranet from outside my house using RDP.

Outside the server on the Intranet.

Hmm.. Thanks for the info. Honestly I think it could be an issue with one of three things:

1. PERC Card (RIP the dream if that's the case)
2. One of the 5TB drives is failing (least likely due to them working in other machines, even in software RAID 5's)
3. Hardware inside the server (ie: backplane, SAS cables, etc.) <---- I find this one to be the most likely culprit since I can no longer create new arrays and my old array is very unstable.

haha PERC card. Yeah i've had my issuess with those myself.

as for failing drive i'm not sure on the impact that would have. I've ran plex temporarely on some really slow laptop drives without problems at all.

As for backplanes they are commonly very stable as the RAID controller/HBA does the work. So my guess is to check the PERC controller.

 

Alternatively find a movie that starts the buffering. Check it on a seperate system (your desktop) as temporarely plex server. Then see if it starts buffering there.

[Joseph Stacklin]

7 minutes ago, AbsoluteFool said:

haha PERC card. Yeah i've had my issuess with those myself.

as for failing drive i'm not sure on the impact that would have. I've ran plex temporarely on some really slow laptop drives without problems at all.

As for backplanes they are commonly very stable as the RAID controller/HBA does the work. So my guess is to check the PERC controller.

 

Alternatively find a movie that starts the buffering. Check it on a seperate system (your desktop) as temporarely plex server. Then see if it starts buffering there.

The only reason I think it could be the backplane is because I jabbed it with a screwdriver trying to get the drive out the first time I got it into my apartment because fedex had damaged the chasis and I couldn't get the drives out because the metal under the right ear was so malformed that i literally had to break the drive free before being able to remove the ear and mold the metal into place. I also may have played linus and dropped my R410 down two stairs twice trying to get it into my server "closet".

 

I'll reseat the PERC card and try to reset the config and maybe even try to throw the firmware back onto it.

[Joseph Stacklin]

Well, it looks like part of the issue at least was with one of the drives. After throwing it into a windows RAID 5 on another machine, Crystal Disk Mark can now read it and it has reallocated sectors and read errors as well as two other Warns. We'll be replacing you as soon as possible.

 

[dalekphalm]

4 hours ago, wolfboytech said:

Well, it looks like part of the issue at least was with one of the drives. After throwing it into a windows RAID 5 on another machine, Crystal Disk Mark can now read it and it has reallocated sectors and read errors as well as two other Warns. We'll be replacing you as soon as possible.

 

Yep 199 bad sectors, 167 of which have been reallocated to spare sectors.

 

That drive is dying. Step one, check the warranty on it. RMA if able - otherwise, nuke it (I recommend DBAN, but there are many choices), then toss the drive (or keep as a coaster/take it apart).

 

Get a replacement ASAP.

[Joseph Stacklin]

4 hours ago, dalekphalm said:

Yep 199 bad sectors, 167 of which have been reallocated to spare sectors.

 

That drive is dying. Step one, check the warranty on it. RMA if able - otherwise, nuke it (I recommend DBAN, but there are many choices), then toss the drive (or keep as a coaster/take it apart).

 

Get a replacement ASAP.

I'll get a new unit soon but have no eta unfortunately.

[AbsoluteFool]

On 14.9.2018 at 3:03 PM, wolfboytech said:

Well, it looks like part of the issue at least was with one of the drives. After throwing it into a windows RAID 5 on another machine, Crystal Disk Mark can now read it and it has reallocated sectors and read errors as well as two other Warns. We'll be replacing you as soon as possible.

 

That's a nice drive to get out some frustration with. Do you have a sledge hammer?