Register for the iXsystems Community to get an ad-free experience and exclusive discounts in our eBay Store.

TrueNAS 12.0-RELEASE Launches!

Western Digital Drives - The Preferred Drives of FreeNAS and TrueNAS CORE

HoneyBadger

Mushroom! Mushroom!
Joined
Feb 6, 2014
Messages
2,758
I'll know in another 14 hours or so how much space to allocate to the metadata and the rest will then go to small files.
Make sure to allow space for data expansion, the usual rules about free space in a vdev still apply, with the bonus that special vdevs stop allocating blocks when they are 75% full (unless this was tuned differently in TN12) and start overflowing to main data vdevs.
 

Constantin

Vampire Pig
Joined
May 19, 2017
Messages
885
Good to know! I plan on being pretty conservative re: metadata since that's the main benefit I get out of 12 - no more waiting for the L2ARC cache to get hot. So I will over-allocate free space in favor of metadata vs. small files. The system is about 30% full, so whatever needs it has now I will likely 5x them just to cover the current HDDs, the 75% fill ratio, and some padding. I expect the metadata needs to be relatively modest after I nuked a bazillion small files by consolidating them into Apple Image files.

But that's the rub - there is no way to know where you stand until you run that zbd command, nor what to look for in the output. So thank you for that insight and I hope iXsystems will devote the resources needed to educate potential users of Special VDEVs re: these sorts of pre-planning tools to make pool transitions and/or creation as seamless and educated as possible.

Winter tires are on now. Still have to verify the respective pressures and reset cranky TPM systems. That's easy... now its time to get dirty. Pressure washing the gaps between boards at the local preschool always leaves me covered from top to bottom with grit and whatever else was trapped in there. Crunchy!
 

Constantin

Vampire Pig
Joined
May 19, 2017
Messages
885
Code:
2.57K   329M   10.3M   82.3M     32K   32.00     0.00      L5 Total
2.57K   329M   10.3M   82.3M     32K   32.00     0.00      L4 Total
3.17K   348M   12.7M    101M     32K   27.49     0.00      L3 Total
98.3K  3.31G    431M   3.22G   33.5K    7.86     0.01      L2 Total
2.91M  94.0G   25.9G    154G   53.0K    3.64     0.53      L1 Total
 442M  14.6T   14.2T   28.1T   65.1K    1.03    99.46      L0 Total
 445M  14.7T   14.2T   28.3T   65.0K    1.03   100.00      Total


So looks like this 14TB pool currently has a need for 94GB of uncompressed metadata, which suggests that my plan for 600GB of metadata space is likely safe since that's >5x the current needs when the pool is 30% full.

Even if we triple that to 300GB for a 100% full pool and then add another 25% that only gets me to 375GB. However, I want to leave some margin in case the future metadata is more bloated than the current set. Hence, 500-600GB seems like a safe bet for my use case (40TB effective pool size after accounting for 80% fill limit and a single Z3 VDEV).
 
Last edited:

HoneyBadger

Mushroom! Mushroom!
Joined
Feb 6, 2014
Messages
2,758
So looks like this 14TB pool currently has a need for 94GB of uncompressed metadata, which suggests that my plan for 600GB of metadata space is likely safe since that's >5x the current needs when the pool is 30% full.
Slightly more. All of the L1 and "higher" objects will be on meta so don't forget about your additional 3.31G+348M+329M+329M = ~4.3G in L2-L5. You've accounted for this for the most part in your tripling of space and acknowledegment of the 80% fill limit, but just to make it known in case someone else is trying to scale it up based on the math.

Also note your ASIZE column for the L1 data is larger than your LSIZE (1.64x) and massively larger than your PSIZE (5.95x) - yet another reason to prefer mirrors for metadata.
 

Monkey_Demon

Member
Joined
Nov 11, 2016
Messages
78
...

And to everyone that suggested I read the release notes - thanks for the advice. I always do - they usually tell me a lot about what used to be broken, but not much about what, if anything, I need to worry about when upgrading. I just thought that since this was being pushed so hard as a pivotal transition point - bringing the code bases together, etc. that there would be more information on the transition from FreeNAS to TrueNAS CORE there might be more information needed. If, as you say, it is mostly a re-branding at this point, then the release notes should probably say that too.
I totally agree with Dean about the need for explicit instructions to transition between FN & TN.

My NAS had been down for about two weeks (network problems), and today when I logged into the Web Interface I was shocked to find "No updates available" in System > Update. A bit further snooping did find TrueNAS 12.0-STABLE.

I was beginning to wonder if I needed to download a new installation file and do a completely virgin installation.

Still, it a concise "how-to" guide would be very helpful.
 

cobrakiller58

Senior Member
Joined
Jan 18, 2017
Messages
397

morganL

Captain Morgan
Administrator
Moderator
iXsystems
Joined
Mar 10, 2018
Messages
261

2bob

Neophyte
Joined
Feb 6, 2018
Messages
7
Bit disappointing that it seems we have to wait until early December to fix the reporting issue given that it worked at RC1 - the 'themes change' workaround doesn't work for me, am i the only one?
 

jlpellet

Member
Joined
Mar 21, 2012
Messages
185
Regarding "Reporting" blank graph but, my workaround still works for me & is only 3 clicks - from dashboard, click Reporting>Services>Dashboard>Reporting & graphs are populated. I don't even look at the screens before clicking.

Glad to have the new link to roadmap & that it does not require logging in to jira.
 

2bob

Neophyte
Joined
Feb 6, 2018
Messages
7
Regarding "Reporting" blank graph but, my workaround still works for me & is only 3 clicks - from dashboard, click Reporting>Services>Dashboard>Reporting & graphs are populated. I don't even look at the screens before clicking.

I've tried that sequence already, it causes the graph axes to become visible but there isn't any data plotted on the graphs
 
Top