YAN (Yet Another Noob) Build

[okgunguy]

After reading that thread, the trouble ticket, and the other link on there, I'm even more convinced that shutting it off is the best thing.
Bidule0hm has the same MB as I do and he has his shut off. I still have all my RAM testing to do so maybe I'll find a problem there as well. But as far as watchdog goes, I'm afraid I'll have to put that dog down.

 

[Bidule0hm]

IIRC you need to do two (three?) things to disable the watchdog, I recall I disabled it with the jumper AND in the BIOS. I don't recall if there was a second jumper to disable but I recommend to check in the manual (read the whole section on the jumpers) just to be sure ;)

 

[okgunguy]

Bidule0hm, I did a 'find' in the Manual for 'watch' and 'jwd1'. There were only two settings entries. One for the jumper and one for the BIOS. No other mentions of it. So according to the manual, it has to be only those two.
Last night I:
- Looked again in IPMI logs to be sure watchdog hadn't reset again, it had not since the last one on 10/11
- powered down (completely) and removed the jumper for watchdog on JWD1
- booted up into BIOS (hit 'delete' while booting)
- checked under BIOS settings for Watchdog setting, it was Disabled. Now this struck me as odd. I'm almost certain that this was Enabled before. The only thing that I could think of was that by removing the jumper, the BIOS saw that it was gone and set it to disabled on it's own. I went ahead and toggled it a few times, put it back to Disabled and Saved and Exited.
- powered down again, plugged in the UBCD flashdrive, removed all RAM, used canned air to blow out all RAM slots and reinstalled one stick
- powered up into UBCD and began MemTest86 - let it run all night
- this morning I pulled it up and it had completed 8 passes with Zero errors
- powered down, moved that RAM stick to the next RAM slot and started MemTest86 again. It is currently running until this evening when I get home and then it will get moved to the 3rd slot... and so on.... I also started a log of what stick in what slot and how many passes with how many errors.
I will update again once I have completed all the RAM testing. I will be trying all possible scenarios/combinations I can think of.

 

[cyberjock]

Ok, gonna explain the watchdog a little...

Watchdogs are a hardware and software solution. The hardware portion keeps a timer while the software resets the timer. It is a hardware solution so that even if the system locks up, crashes, etc you still have control of the hardware. The software (watchdogd to be precise) should reset the timer at regular intervals (let's say 30 seconds) and the timer is set for say 5 minutes.

If the system hasn't crashed or locked up, the timer should reset before it ever counts down to zero. Locked up typically means you have no ability to troubleshoot further anyway, so choosing to disable it because "I want to troubleshoot a problem" is a lost argument. If it crashes and you aren't a kernel debugger, again you have no reason not to reboot. You may want it to reboot anyway so that some kind of trigger in some clustered arrangements has another node pickup the load. The system may also recover after the reboot. Ideally, this is what "should" happen if something totally unexpected happens. You should be able to look at the historical logs as well as reporting data to put together a plan of what may have happened and then begin troubleshooting further. In any case, a proper setup with proper logging, etc shouldn't typically require you to turn off the watchdog timer unless the timer is actually broken.

It's typically set to 5 minutes instead of 1 minute because from POST to OS bootup3 minutes may pass. You don't want the system rebooting during the bootup sequence because you made the timer too short, do you? ;) Many can be configured for appropriate timeframes and it is important to recognize what is appropriate for YOUR setup. Copying someone else's value because it works for them is stupid. You must understand the tech (I've been saying that alot lately).

Now, FreeNAS only supports a small number of watchdogs (some of the most popular). If its not supported, you'll get those pesky reboots at regular intervals because nothing is resetting the watchdog timer and the timer keeps on tripping without being reset.

In this case, you're probably better off with it disabled if you are getting random reboots because your watchdog may not be compatible with FreeNAS' watchdogd.

 

[okgunguy]

Thanks for explaining that so thoroughly.
And for calling me stupid. :( <- That's me giving you the crazy eye!

 

[okgunguy]

While I wait on my RAM to finish testing. I had a quick question I wanted to pose for whomever would like to answer it.
How do most users have their Volumes and Datasets set up? Currently I'm thinking about setting up one dataset for Backups and one for Media. And that's it for now. Then inside Media I'll have my folders for Movies,TV,Pictures and Music.
Do most users have it set up like this? Or do you have a separate Dataset for each type of media? I've seen the "instructional videos" showing to do it both ways.
If you do have separate Datasets for Movies,Music, Tv.. etc. Why do it that way? Do the other jails (couchpotato, SickRage, Sabnzdb) need them separated out like that? Or will it all work fine in folders under one dataset?
I'm just trying to keep it as clean as possible. But I can't seem to find any definitive answers to why one config is better than the other.

 

[depasseg]

The jails (and sharing services) don't care about the datasets, you could have one or 50.

I'm sure others have different reasons, but to me, one reason to have different datasets is to allow different sharing protocol (have one dataset for CIFS and a separate one for NFS, for instance). Also, since snapshots and replication are done per dataset, think about the snapshot frequency, retention and replication recovery requirements.

Here's mine.

 

[Nick2253]

Also, since snapshots and replication are done per dataset, think about the snapshot frequency, retention and replication recovery requirements.

This is the big reason I have different datasets for different kinds of media.

For example, I rarely go back and watch TV shows once I go through them (I'm a binge watcher), so I really don't care that much about snapshot frequency. But movies, on the other hand, I go back to regularly, so data corruption and the like would be inconvenient (I'd have to re-rip a bunch of stuff). Hence, my snapshot schedule and retention is very different for the two kinds of media. But that's just me.

 

[Robert Trevellyan]

How do most users have their Volumes and Datasets set up?

To generalize what others have said, there are settings and properties and capabilities of datasets that are not present with sub-folders. If you can see a potential reason to do those things differently for a particular subset of your data, that's a good reason to put that data in its own dataset. If in doubt, I would lean to more datasets rather than fewer, for flexibility.

 

[okgunguy]

I am no where near understanding all the parameters of setting up the other jails (couchpotato,sabnzbd,sickrage,etc) so I don't know if they prefer separated datasets or if they like the sub-folders. I am leaning more towards separated because of how ya'll have pointed out the snapshots and the shares. I like that I could take snapshots more frequently of my movies and keep PC backups for longer and can restore them all separately if need be.
I just want to get it right for the future if/when I get the automated portion figured out. I'm half tempted to set all of the automated programs up on another PC (my refurb'd/interface pc) and keep it completely off of my NAS. But it doesn't have near the power my NAS has.
My RAM testing is almost complete, so I need to decide. Sounds like I'll be doing more reading on the jails. Thank you all for all the sage advise. It is always appreciated.

 

[diedrichg]

To make it so much easier to get SAB, CP, Sickrage, Sonarr, etc to talk to each other, it is recommended to manually install them all in one jail rather than using the plugins. This also gives you the power to upgrade those applications at will.

 

[okgunguy]

RAM testing update: 4 sticks of 8Gb each / all HDs unplugged
Almost complete. All single sticks in each slot have passed. I let them run for 8-10 hours each test. Running average 8 passes each time. Zero errors.
I ran into an abnormality (or maybe it's not) on testing 2 sticks at a time. When I had (only) 2 sticks in Dim A1 & B1, I let it run for 24 hours. Ran 13 passes, Zero errors. Then when I ran the other (only) 2 sticks in Dim A2 & B2, I let it run for 24 hours but it only had 8 passes. Zero errors. Why slots 2 only got out 8 passes for the same amount of time is the anomaly. Can only think of my MB has that thing where it needs slots 1 filled before it can use slots 2?? Don't know the terminology to explain it properly.
Currently have all 4 sticks in. Probably will let it run for 48 hours. Then another 48 with the drives plugged in. Then I'm going to have to call it good. Unless my anomaly is a cause for worry. Then I have worries.

 

[nightshade00013]

I don't think there is a lot to worry about on the number of passes in a particular timeframe. I got my final two sticks in and wanted to run a couple passes on just them alone to verify no problems on just 8GB rather than running all 48GB in one shot. First pass completed in 45 mins, so second one should be done in another 45 right? Wrong, second pass took about 75 mins. Some sticks will process a little better than others and as they warm up they can slow down just a bit. My first test was a fresh boot from cold with two sticks that had just showed up in the mail, after they ran for a while they warmed up and slowed down a little.

 

[okgunguy]

I have put my NAS back to operational state. I'm currently searching through trying to find the most comprehensive and easy to understand guide to installing Plex. Since I'm not trusting the way I installed it last time. Last time I followed along with Ixsystems YouTube video which instructed me to do everything through the GUI. Which I'm assuming is as a PlugIn??
I'm not really understanding what the differences are between Jails and Plugins. I know all the BS answers that you can update Jails whenever you want and have to wait for Plugins to come out with a update. But nobody has given any real description on why Jails are better. Are they more stable? Are you able to configure them more in-depth than Plugins?
I'm currently leaning more toward a thread written by Cosmo_Kramer titled (Tutorial) How To Install Plex in a FreeNAS 9.3 Jail (Updated) it isn't making much sense to me but I guess once I start doing it maybe it will. I am not a big fan of just typing commands without knowing exactly why I am typing them. But it is about the only tutorial I've found that is written for 9.3.
I've still got more testing to do on my system before I go installing anything on it.
So if anyone has a better/different tutorial on Plex they'd like to share I'd love to read up on it.

 

[nightshade00013]

I just do it though the Jail shell interface. Create a jail and name it whatever you want. Then select the jail and at the bottom click the shell icon(second from the right big black looks like a prompt)

pkg update
pkg upgrade
pkg install plexmediaserver

Once that is all done and you follow all the prompts you can

echo plexmediaserver_enable="YES" >> /etc/rc.conf

service plexmediaserver start

Add some storage to the jail and then look at the jail's IP address in the jail tab, go to it and make sure to add /web to it.

The rest of the tutorial is dealing with permissions if you want plex to have the ability to delete files. Have not gotten it to work right as of yet since they files are also a part of a windows share but I am also not worried if plex can not delete a file. I can always do that other ways anyway.

So the basic setup is only five steps, update the pkg system, upgrade the pkg system and any other software that needs it, install Plex, add plex to /etc/rc.conf, and finally start the service.


jexec is nice if you want to use ssh to the main box and do a lot of file editing but it's pretty easy to do other ways and take a couple steps out and just get the basics going.

 

[Robert Trevellyan]

I'm not really understanding what the differences are between Jails and Plugins.

A standard jail is like a blank slate that you can do whatever you want with. A plugin is a preconfigured jail designed to serve a specific purpose.

But nobody has given any real description on why Jails are better.

There is no "better" without context. Using a jail gives you more flexibility, including the ability to install multiple applications in one jail, but you typically have to do more work to get the application you want up and running. A plugin is more of a turnkey solution for a specific application, but you're stuck with the decisions made by the plugin's author, and it's one application per jail. For an analogy, FreeNAS is to FreeBSD as plugins are to jails.

My current preference is to run applications in VirtualBox VMs, but that's mostly because I'm so much more familiar with Linux than I am with BSD. When I need to figure something out, my google-fu works better for Linux than for BSD. EDIT: this approach has a significant downside, which is that a VM works ZFS much harder than a jail does.

 

[okgunguy]

Robert, that is the best description of the two that I've seen yet. Thank you for clearing that up for me.
I've got a quick command question. I'm in the process of configuring my datasets and I have some directories to move. I have my Movies directory in my main dataset and would like to move it to my new Media dataset.
Would - mv /mnt/r2d2/Movies /mnt/r2d2/media - move "Movies" into "media"?
Or would I need to do - mv /mnt/r2d2/Movies /mnt/r2d2/media/Movies - ?? Or do I need a * in there somewhere to get all the files as well? I don't want just a copy of it, I want it completely moved.

 

[rogerh]

Robert, that is the best description of the two that I've seen yet. Thank you for clearing that up for me.
I've got a quick command question. I'm in the process of configuring my datasets and I have some directories to move. I have my Movies directory in my main dataset and would like to move it to my new Media dataset.
Would - mv /mnt/r2d2/Movies /mnt/r2d2/media - move "Movies" into "media"?
Or would I need to do - mv /mnt/r2d2/Movies /mnt/r2d2/media/Movies - ?? Or do I need a * in there somewhere to get all the files as well? I don't want just a copy of it, I want it completely moved.

As an only occasional CLI user, my first resort on reading your question and realising I was not sure of the answer was to read 'man mv' - the conclusion I reach is that either of your proposed commands would have the same effect, which is the one you want. I could be wrong, but I recommend the man pages for the exact syntax of commands. Edit: I am not sure now - maybe your second one will give you /mnt/r2d2/media/Movies/Movies which is what you don't want; I don't think so though, as long as /mnt/r2d2/media/Movies doesn't yet exist.

 

[Robert Trevellyan]

Would - mv /mnt/r2d2/Movies /mnt/r2d2/media - move "Movies" into "media"?

Yes, it would.

would I need to do - mv /mnt/r2d2/Movies /mnt/r2d2/media/Movies - ??

You'd get an error with this version, assuming the 2nd folder doesn't exist. From the man page:

SYNOPSIS
mv [-f | -i | -n] [-hv] source target
mv [-f | -i | -n] [-v] source ... directory

DESCRIPTION
In its first form, the mv utility renames the file named by the source
operand to the destination path named by the target operand. This form
is assumed when the last operand does not name an already existing direc-
tory.

In its second form, mv moves each file named by a source operand to a
destination file in the existing directory named by the directory oper-
and. The destination path for each operand is the pathname produced by
the concatenation of the last operand, a slash, and the final pathname
component of the named file.

You're not renaming a file, so you need to use the 2nd form.

I recommend the man pages for the exact syntax of commands.

+1, and try things out in a sandbox if necessary.

EDITed for accuracy.

 

[rogerh]

Yes, it would.

You'd get an error with this version, assuming the 2nd folder doesn't exist. From the man page:

You're not renaming a file, so you need to use the 2nd form.

+1, and try things out in a sandbox if necessary.

EDITed for accuracy.

OK I accepted the implied challenge: the "first form" in the man page, where you name the nonexistent directory you want does work because, a) mv seems to treat a directory (and its contents) just the same as a file, and b) you can "rename" it to the same name as it originally had and mv doesn't care. Tested in an OSX terminal, but I doubt if FreeBSD mv is different.