More unRAID New User Questions

[AgentXXL]

First off, let me thank forum members who've already replied to my previous questions: Squid, bonienl, itimpi and dlandon have all been great helpers for me and my migration from my mostly reliable old FreeNAS setup to unRAID. I will be donating to each of these people (and others) as soon as I get my next disability cheque later this month. Of course I will also be purchasing my unRAID Pro license then too, as I'm only 11 days into my trial. 

 

Second, sorry for this LONG post, but rather than litter the forums with multiple posts, I'll try to detail my current concerns and questions here. Feel free to respond to just the items you're comfortable with. That's one of my concerns, the fact that searching for help here, on Reddit, and elsewhere often finds posts/comments for long-since deprecated versions of unRAID or the associated plugins I'm using. This already led to to at least one issue where I tried to implement a solution that was no longer valid for unRAID v6.x. See item #5 below for more on this.

 

I've made great progress, specifically in migrating a lot of my media content to the array, but the more I copy to it, the slower it seems to be getting. It's currently at 26TB with 2 new 8TB drives (one for parity, one for data), a lightly used 6TB drive, and two older drives, one 4TB and one 8TB. The older drives were stress-tested using manufacturer tools and then pre-cleared under unRAID before being added to the array. I'm also using a single 1TB SSD for my cache drive, but plan to implement a 2nd 1TB SSD for the cache pool soon, for redundancy. 

 

I have an LSI 9201-16i with cabling on the way so right now I can't  expand the storage further without replacing drives and letting them rebuild. Once the controller/cables arrive, I'll be able to add more to my 20 bay hot-swap SATA chassis (Norcotek RPC-4220). However, I'm concerned with a couple of things:

 

1. Performance of the system isn't great for copy/move speeds, as I suspect the parity drive speed is part of the issue. It's currently a new but stress-tested Seagate ST8000DM004 shucked from its USB enclosure. I need to upgrade the drive to a larger size so I can eventually shuck the 5 x 10TB drives I purchased a few months ago. I'm planning to go with at least a 10TB but considering a 12, 14 or maybe even one of the new 16TB drives. Is there a big benefit from using an Ironwolf Pro over the regular Ironwolf, or the current Barracuda? And yes, I'm fine with using WD/Hitachi drives too... I've used and experienced failures from all drive manufacturers over the last 35+ years of my computing life, but the Seagate's are usually easier to find on sale here in Canada. I'm suspecting a large (and also very expensive) 10TB+ SSD isn't a good choice due to the high number of writes to the parity drive.

 

2. I have my user shares set to Yes for Use Cache Drive. While I like how this works, I also get frustrated by the CONSTANT reminders that the cache drive is over the watermark capacities set in Cache and Disk Settings. If you choose Yes, and want data to use the cache drive 1st, then write to the array once full, why can't it ignore the levels set in Disk Settings? I've set the disk utilization threshold % for warning and critical levels to 95% and 99% respectively in the Cache settings (pic below). Alas it seems to ignore these and use the main levels configured under Disk Settings. I could change the levels in Disk Settings to the same, but I prefer not to as that affects the array. I know this will continue to be an issue as I still have 50TB of data from USB and the old FreeNAS to migrate into the array.

 

3. Mover performance is horrible. While using the Cache drive 1st, then writing to the array once full, I end up with about 900GB of data to move on the next scheduled Mover run. Or even if invoked manually from the Main tab of the unRAID webGUI, Mover takes a VERY long time to move this data to the parity protected array. I still want Docker and VM data to reside on the SSD cache drive, including data that the Dockers/VM create/download. I suspect if I set Use Cache Drive to No, that would prevent my Docker containers and VM (not yet implemented, but planned shortly) from creating/downloading data to the cache SSD. Or is there a way to ensure that the cache drive is used for Docker/VM data, but not for importing large amounts of data from my Unassigned Devices USB drives and network copies?

 

4. Performance and tuneables... as much as I've tried to read about these settings, I'm still unsure what to set them at. My system is based on an Asus Maximus Gene VIII motherboard, an i7-6700K cpu, and 32GB of RAM. Older components but as I'm on disability, it's what I have to use until I can save up enough money to put together some new hardware. I realize a lot of these performance issues are because of the 50TB or so of content I need to migrate from USB/network locations to the array, and once all on the array, the day-to-day performance of unRAID will be better. But in the meantime, I'm still finding it horribly slow to copy data to the array. Again, I know that parity calculation is part of the reason as mentioned in item #1. Can someone recommend some decent settings for the tuneable settings shown below?

 

5. The horrid state of the old and often deprecated information in the unRAID forums, wiki(s), Reddit, etc is HIGHLY confusing to those new to unRAID like myself. I have the official documentation and try to cross-reference it when searching for solutions, but there are SO MANY older posts that still show up in searches, even when I add 'unRAID 6' to try and limit my searches to current info. This is unRAID's biggest drawback for new users - too much old and conflicting info that is no longer relevant. I appreciate that these forums have some separation for v6.x and older versions, but web searches still often give me results that are no longer valid. Not much that can be done, but a shorter, more concise FAQ/tutorial for new users is something that I think is sorely needed.

 

Any help is appreciated. Again, sorry for the very long post, but hopefully I've provided enough info to make it easier to answer some of my questions. I can upload diagnostics if requested. Thanks in advance for any suggestions/assistance!

 

Dale

 

 

[Squid]

50 minutes ago, AgentXXL said:

Performance of the system isn't great for copy/move speeds,

You can enable Reconstruct Write (Turbo Mode) in Disk Settings, but for the most part, I get in and around 70-80MB/s writes over the network to the array, and 100+ MB/s to a cache-enabled share (with it set to auto)  On a 1G network, you're limited to ~110MB/s maximum, so at the end of the day I don't have any major problems with the speeds I get

 

50 minutes ago, AgentXXL said:

I have my user shares set to Yes for Use Cache Drive. While I like how this works, I also get frustrated by the CONSTANT reminders that the cache drive is over the watermark capacities set in Cache and Disk Settings. If you choose Yes, and want data to use the cache drive 1st, then write to the array once full, why can't it ignore the levels set in Disk Settings? I've set the disk utilization threshold % for warning and critical levels to 95% and 99% respectively in the Cache settings (pic below). Alas it seems to ignore these and use the main levels configured under Disk Settings. I could change the levels in Disk Settings to the same, but I prefer not to as that affects the array. I know this will continue to be an issue as I still have 50TB of data from USB and the old FreeNAS to migrate into the array.

You can install the Mover Tuning plugin to allow you to fine tune the settings for when files get moved from the cache drive to the array.  Personally though, most of my shares are set to not use the cache, as the speeds I get (see above) are good enough.  I only enable them for when I really care about transfer speeds, or for the appdata.  IE:   My media creation is all handled automatically by various containers, and I don't notice (nor would I care) that a transfer from a download share to the array takes a minute longer.

50 minutes ago, AgentXXL said:

Mover performance is horrible.

That's because it has to perform multiple checks to see if the file is in use, etc.  No way around that.  Once the move on the file begins though, it's going to work at the maximum speed allowed.  But, if you're moving tons and tons of small files (.nfo files eg), then the overhead becomes quite extreme.  Another reason to not bother with enabling use cache for media shares.

50 minutes ago, AgentXXL said:

Can someone recommend some decent settings for the tuneable settings

For the initial data load, enable as wd_write_method Reconstruct write.  Much faster at the expense of every drive having to spin, but you will never be able to exceed the speed of the network regardless of what drives you've got installed.  (Assumes of course you're transfering directly to the array instead of the cache drive.)  For the other settings, the easiest check is what is your parity check speed after it runs for around 10 minutes with no other activity happening on the array.

50 minutes ago, AgentXXL said:

wiki(s),

The wiki sucks.  Best to completely forget it ever exists, unless someone points you to a particular entry in it.

50 minutes ago, AgentXXL said:

Reddit

While occasionally I do go to the reddit page (and the facebook page), I am actually always surprised that people keep asking questions there with problems.  IMO, if you want the most reliable answer to any given question, then you're best off here.  Not to say that any given answer on reddit or facebook is incorrect though.

50 minutes ago, AgentXXL said:

SO MANY older posts that still show up in searches

Welcome to the internet.  Everything shows up that's ever been posted for ever.  I keep expecting for my kids to sue me in later life when they discover some pictures I posted of them when they were babies.

50 minutes ago, AgentXXL said:

FAQ

Several FAQ's exist here on the forum.

 

50 minutes ago, AgentXXL said:

tutorial

Check out @SpaceInvaderOne's videos

 

Edited June 9, 2019 by Squid

[AgentXXL]

15 minutes ago, Squid said:

You can enable Reconstruct Write (Turbo Mode) in Disk Settings, but for the most part, I get in and around 70-80MB/s writes over the network to the array, and 100+ MB/s to a cache-enabled share (with it set to auto)  On a 1G network, you're limited to ~110MB/s maximum, so at the end of the day I don't have any major problems with the speeds I get

Should this be done just for the initial load of my 50TB+ that are on USB attached (UD mounted) drives? Or should it be left at this setting permanently? I'm at 1G networking but do have a better switch on the way as my 'prosumer' gear is dated and not as efficient as I'd like. The data copies from my old FreeNAS are usually decent speeds of 80 - 110MB/s, but I'm seeing packet loss that occasionally drops the speed to 20MB/s or less. Which is why I'm getting the new (used) switch.

 

23 minutes ago, Squid said:

You can install the Mover Tuning plugin to allow you to fine tune the settings for when files get moved from the cache drive to the array.  Personally though, most of my shares are set to not use the cache, as the speeds I get (see above) are good enough.  I only enable them for when I really care about transfer speeds, or for the appdata.  IE:   My media creation is all handled automatically by various containers, and I don't notice (nor would I care) that a transfer from a download share to the array takes a minute longer.

So you have your shares set to No for the cache drive? That's what I'm thinking as well... if I can get the USB and network speeds to stay above 50MB/s then I have no concerns of not using the cache drive for my shares.

 

28 minutes ago, Squid said:

Several FAQ's exist here on the forum.

 

Check out @SpaceInvaderOne's videos

I'm pretty much in agreement that the abundance of old info on the internet is the bigger issue. That's why there should be a clearer tutorial/guide for new users. Certainly that's more an issue for the LimeTech team, but it also wouldn't hurt if we made a "NEW USERS START HERE" forum that gets updated with pointers to trusted FAQs for the current releases. I'm willing to help once I have more familiarity with unRAID. My issue with the FAQs is that they aren't always updated regularly enough to help out new users like myself. I will try and take a look at the videos you mentioned. 

 

Thanks again Squid!

 

[Squid]

1 minute ago, AgentXXL said:

drops the speed to 20MB/s or less.

Something wrong somewhere.  But, I do have to say that unless I've picked up a switch at a dollar store then I've only very rarely seen any significant packet loss.  And even the odd packet loss is OK.  Most users who've questioned here about seeing packet loss shows, the statistics show that its < 0.0001% which is basically nothing (and bound to happen at times) (My current packet loss is 0.000000000175%)

 

[Squid]

10 minutes ago, AgentXXL said:

"NEW USERS START HERE"

https://unraid.net/blog/unraid-new-users-blog-series

[AgentXXL]

2 minutes ago, Squid said:

https://unraid.net/blog/unraid-new-users-blog-series

PERFECT! Now if it would only be prominently displayed and/or linked to via the main unRAID docs, then new users like myself won't have to struggle finding out what content is bad/old. I'm already feeling much more comfortable with unRAID, but I'll still read through it. Thanks!

 

 

[Squid]

@SpencerJ

[AgentXXL]

@Squid I went ahead and set my shares so that the cache pool is no longer used and data is written directly to the array. My Docker apps for creating and downloading content still make use of the cache pool (SSD based) so I get reasonable performance. However, I'm noticing a lot of lag with both the main unRAID webGUI and for the webGUI interfaces of many of the Docker apps.

 

I suspected my network and the random packet loss I previously mentioned might be at fault, but my monitoring software isn't seeing much that's network related. I do however note that since enabling the Reconstruct Write mode, my CPU usage is much higher. It's a 4 core (8 threads with hyper-threading) i7-6700K, not overclocked, but water-cooled. I've seen numerous instances of multiple cores/threads at 100% utilization. Perhaps the Reconstruct Write method is too much for my CPU? Thoughts?

 

Dale

[SpencerJ]

22 hours ago, AgentXXL said:

PERFECT! Now if it would only be prominently displayed and/or linked to via the main unRAID docs, then new users like myself won't have to struggle finding out what content is bad/old. I'm already feeling much more comfortable with unRAID, but I'll still read through it. Thanks!

 

 

Hello,

Are you referring to our wiki?

 

Thanks

[AgentXXL]

Just now, SpencerJ said:

Hello,

Are you referring to our wiki?

 

Thanks

To the blog for new unRAID users... it should be featured more prominently so users new to unRAID don't spend countless hours weeding through old and often out-of-date info from most web searches. I'm very computer literate, but only 12 days into unRAID and it's been one of the most confusing to get current, reliable info for new users. These forums and users mentioned above have been of great assistance, and after reading through the blog post @Squid mentioned, it's something that should be a 1st stop for all users new to unRAID.

[SpencerJ]

12 minutes ago, AgentXXL said:

To the blog for new unRAID users... it should be featured more prominently so users new to unRAID don't spend countless hours weeding through old and often out-of-date info from most web searches. I'm very computer literate, but only 12 days into unRAID and it's been one of the most confusing to get current, reliable info for new users. These forums and users mentioned above have been of great assistance, and after reading through the blog post @Squid mentioned, it's something that should be a 1st stop for all users new to unRAID.

Thank you for the feedback. I've enabled the date to show up in the Hero copy of the new users blog (publish date was listed in the https://unraid.net/blog page but I have also enabled it on the blog page itself and will continue to do so in the future) and all New Users related blogs will have the "New Users Series" tag. These tags are sortable from the main blog page.

 

As for old information, we can only control so much with regard to google searches. FWIW- we are in the process of attempting to make the wiki more up-to-date and streamlined but this is a very time-consuming process. Agreed with @Squid re: "IMO, if you want the most reliable answer to any given question, then you're best off here.  Not to say that any given answer on reddit or facebook is incorrect though."

 

Thank you again for the feedback. If you ever have any suggestions for new user blogs or information you would like to see written about w/r/t Unraid, please don't hesitate to reach out to me here via DM or email ([email protected]). 

 

Cheers