Search results

The memory requirement of the L2ARC depends on the block size of the cached data. IIRC each L2ARC chuck requires 180 bytes of memory, and by default a maximum of 25% of RAM can be used for the L2ARC index. I made a basic calculator in Excel a few years back using the numbers I found in some...

"zfs import" Thats the "level of skill" you need... Who wants to be tied to a single manufacturer that might or might not be on the market when I need a new controller that *will* eventually fail? What if they decided that they no longer need to support arrays made on legacy controllers...

Not if you use public links. People can just download the file from the web.

The price of Onedrive is hard to beat, even more so if you have any use for their Office Suite. The family plan gives you up to 5x 1TB at a very reasonable price.

St3F Did you follow a specific guide to installing ZoL on Xubuntu? I tried getting ZoL to work on Ubuntu not long ago, without much success...

Wow, high level thread necromancy, buddy!

I don't think it's ever even been mentioned by anyone, let alone announced by Samsung. No one knows if there'll even be an 860 model at some point. Why would there, anyway? Samsung seems to have changed their naming with their new form factor SSD's, and I don't see much reason to release new...

You blame Samsung for something that hasn't happened? What the actual...? Sure, Samsung released firmware that had a bug (which is effectively impossible to detect in Q&A due to the time it takes for it to show up). Then, when they postpone the fix to make sure it's as perfect as they can make...

Yes, and in these situations you use spinners (typically 10k) and SSD or flash caches. You would never, ever, use 20k drives. The fact that you think otherwise - and insist on it - tells me that you haven't set foot in a DC in the last 5 years. Even cheapish servers from the major players...

I don't mention heat or power at all, I mention R&D (basically developing a completely new type of harddrive) and manufacturing (something like this is going to require some mental tolerances), and if you want it to sell, it has to be WAY cheaper/GB than SSD. You'd have to be one hell of a...

Even if you happened to be correct, it's not being wise to just wildly guess and then lucking out. It's called blind luck. There are no indications that any manufacturer is making a 20k+ RPM drive, and with good reason. The cost in R&D alone is ridiculous, manufacturing it, probably just as...

Whaaat? I think you're the only person on the planet that thinks this.

Okay, I'm throwing in the towel.. I need help! I'm trying to move the metadata of my PMS to an nfs exported ZFS filesystem. It all seems to work fine: I can mount the exported filesystem and I've copied the old PMS metadata to the mounted filesystem. Now PMS won't start, and I think the...

I did the same a while ago.. It'll rebalance over time, so if the speed discrepancy is a non-issue to you, I would leave it as it is..

If you think PCI-X will work in PCI-E, I don't think you're quite ready to play around with Fibre Channel.

I love how JoseJones is running a system from 2009 that seemingly hasn't had a single upgrade since then, and all of a sudden it's like SOOOOO important that he can upgrade from super fast SSD to SUPER DUPER fast SSD every year or so because he's gonna feel the difference between 0.05 ms and...

If you have 48 GB of RAM in the machine, you'll have a huge ARC. I doubt you'd ever see any hits on your L2ARC devices, so they're basically just wasting ARC for the L2ARC table. I would either ditch the L2ARC and monitor the ARC hits - if you have lots of misses, an L2ARC device might make...

Wow, what's with the attitude, zambanini? We don't need or want that around here.

It's not the backplane that can cause issues, it's the expander. And you're correct, OP, there have been widespread reports of issues when people use SATA drives on SAS expanders. I'm fairly sure you can find lots of info on it in the SAS expander owners thread in this subforum.

I can't wait for the day when people stop making purchasing decisions based on personal anecdotal evidence. It's frustrating, yes. But unless you actually believe that your sample size of two (2) drives is enough to warrant an outburst like this, you're just spreading FUD. And if you do think...

Pass the controller that has the disks through to the VM and import the pool. Do not let ESXi touch the disks in any way.

I wouldn't expect any issues using the MX100 for this task.

Why don't you want to pass through the controller?

How can this be your recommendation for "cheap/decent" ? It's so expensive that most enthusiast skip it for it's cheaper siblings like the 840 EVO or MX100. My personal recommendation would probably be the MX100 due to the recent firmware bugs with the 840 EVO.

Yea, you need to remember that backups are important. It's not just some afterthought like "hey, I've got some old not-yet-dead drives that I can pool in an old impossible-to-get-spare-parts-for server! Lets play backup service provider!" Your customers will probably expect you to have at least...

ECC RAM is no more or less important using ZFS than any other filesystem and you do not need 8 GB of RAM. ECC RAM will increase the reliability, yes, the same as with any filesystem. More RAM will increase performance, yes, but not for mostly static data, which I assume this is. And seeing as...

Backups on RAID 0? Remember, if you have ANY versioning implemented in your backup system, your backup system holds primary data. If your backup data is gone, those older versions are gone. Do not use RAID 0 for your backups if you need or want older versions of your data.

No, Napp-it adds a GUI to a free OS that has enterprise features. You get all three.

You've got something completely backwards regarding GUI vs. CLI. Even Microsoft is switching back to CLI for their server OS'. You're in the classic pick-two-out-of-three situation: Enterprise features GUI Free Pick any two. (napp-it being the only option that comes to mind that delivers all...

Why does this sound revolutionary to you? Fewer disks = less floor space required. Floor space is part of the $/TB equation in any data center. Some applications only care about capacity. When you're buying HDD's a few racks at a time, putting your storage pools behind storage virtualization...

This is completely correct. You can never improve your ZIL performance by adding SLOG devices. Even if you had a stripe of 10 drives, the ZIL would never start writing the second write IO before the first one is completely written to disk. There is absolutely no parallelism here.

I'm not saying SSD's are expensive. Neither is RAM. Neither is upgrading to a faster CPU. And getting a redundant PSU isn't too bad either. Hey, you can get a 10 gbit NIC off of ebay quite cheap as well. And why not get a hotswap backplane? That's only like 80 bucks! My point is - nothing is...

30 seconds? The ZIL is written every 5 seconds unless under very intensive load IIRC. Regarding the "freaking out and shutting down" I have no knowledge of this. Do you have a source for your information? Bug report perhaps? I'm not disputing the fact that mirroring your SLOG is best...

I have to disagree with a lot bds1904 is saying here. For most, 32 GB will be fine. It obviously depends on the number of virtual machines and their workload. Ask yourself what you want to use the AIO for and base your build on that. If a 32GB maximum is fine, then that board and CPU combo will...

I'm fairly sure the red line isn't "used" as in "in use by the operating system" but what ESXi considers active. In use does not equal active.

I don't believe I stated anywhere that it was, but I guess I could've been more clear. Thanks for clarifying!

Lots of people on this forum are using the M1015, myself included. It's the go-to HBA for a few good reasons: It's well supported It's cheap It's simple (no BBU, no cache, no XOR processor) Regarding ECC; it can save your data. You mention that spending a bit more money to increase your RAM is...

ECC doesn't magically enable any ZFS features. It's an extra level of safety which should be used on any file server where the data is important, regardless of which OS/file system is deployed. Also, the 1GB RAM/1TB data thing needs to die. This is not a requirement. It'll give you more ARC...

I stand corrected! This is awesome, although not very useful in most use cases. Thanks for the info!

You can't do 1 mirror with 4 disks, it'll always be 2 vdevs with 2 disks each (unless you're doing 3-way mirrors of course)