RobJ

Could you give a reference for your understanding that user shares are already managed in memory? The User Share file system is a virtual file system built by FUSE in memory, which should make it unnecessary (and wasteful and inefficient) to require caching it in additional memory. I've always felt that CacheDirs is not useful at all for User Shares, but I vaguely remember a very knowledgeable user saying there WAS a use in certain circumstances, which I don't remember. I'd say uninstall CacheDirs and see if you ever spin spun-down drives up while browsing the shares.

I've previously requested ata info here. Unfortunately, it's not invariant, it can change as much as the sdX assignments. However, in a small system with few controllers it can be consistent enough to appear invariant. The more controllers you have, the more the order of loading them can be different.

Fireball3 has done some nice wiki work on other things. Perhaps he might be willing to set up a wiki page based on this thread, with different sections for each card, and copy the appropriate instructions and links to them?

Drive looks fine. With 180 pages in this thread, you can be excused for not reading it all! We've often commented about the error rates, and their ridiculous numbers on Seagates (unlike other brands which don't use the RAW here), but haven't mentioned it in awhile, so might as well again. The Raw_Read_Error_Rate and Seek_Error_Rate are just error rates, not counters, and their RAW values are actually coded fields, but displayed (wrongly) as a decimal number. If you were to examine that number in hex across multiple SMART reports, you would be able to see parts of the hex string changing, but much of it not changing. Displaying it as a decimal is always misleading. More importantly, its construction is only of use to the manufacturer, not to us, so it should always be ignored. What is important with error rates is how the VALUE and WORST are behaving, whether they are dropping, getting close to the THRESHOLD. Because they are based on rates, not counters, they jump around some, and can go up and down. Usually, 100 (or sometimes 200) is set to correspond to an optimal rate of read or seek errors, but at times the rate may be even better, producing numbers like the 119 you currently have. There are exceptions in how they are handled though. For example, Seagate optimal seek error rates are typically in the 60's, not near 100. Hardware_ECC_Recovered is just a copy of Raw_Read_Error_Rate, ignore it. The 2 that ARE a bit unusual (but only minor items to monitor over time), are High_Fly_Writes and Command_Timeout, particularly because it's a brand new drive. High_Fly_Writes is 19 already, can sometimes indicate mechanical issues in the future. Just monitor it. The scaling function for the VALUE is a dummy routine, ignore what happens to it. Command_Timeout is 1503, somewhat unusual, but probably not a problem, especially since VALUE hasn't budged from 100. My research found two rather different (and both confusing) descriptions for it, and nothing yet that is conclusive about it. As usual(!), different vendors are using it quite differently. I personally have seen no correlation with any other issue at all.

Yours looks essentially perfect too, and with a Raw_Read_Error_Rate of 119, you could say it's better than perfect!

This is not the first time comments like this have been made, and I shudder every time! Yes, it would be great for unRAID and LimeTech, but then I think of the accompanying support nightmares! Trying to support non-technical users is HARD(!), and costly! I remember someone saying something like, "if X was easier, then LimeTech could make an extra 25 million!". And it's possible that's true, but I personally believe that gain would be completely or mostly eaten up by the support costs. Support is generally the most expensive part of software development, I believe. When I look at the General Support board, I get overwhelmed very quickly (I overwhelm easily!). With the increase in users lately, there is also a big increase in support requests. I do appreciate however the increase in helpful people there, providing good help too! A good thing too, because some of us older helpers aren't there as much as we used to be. I'd love to help, but find myself almost avoiding that board because of the volume. I'd like to see some stratification of the infra-structure, those who like doing first-line support (welcoming, reassuring, dealing with the simple problems, requesting the needed info, categorizing and passing on the harder issues), and the more specialized helpers who deal with the different categories of harder issues. I have no idea how to make it happen though. These are great ideas, probably the kind of thing that happens as a product matures. We're at that point I think, and UI and infrastructure standards would be a good thing to begin building. I do think that LimeTech developers (and I especially include bonienl in that list) have the primary lead here, because the one thing that helps the most in UI consistency and style is the tools and guidance provided by them (think 'Material Design'). If the current Help system was made easily available to plugin developers, that would go a long way towards making the Help system more universally used. If a little question mark was wanted, for either click-on or hover-over help, then they would need to provide the tool, in a way that plugin authors (and perhaps even third party writers) could easily and consistently use. As an example, Squid's work (templates, etc) has been forcing consistency on container preparers. I particularly hope that third party help would be possible, as none of us are universally gifted. Those who are great in authoring plugins and dockers are not necessarily good at the help needed by beginners. Others here would be much better at adding the hand-holding and readable detail needed, guided by the original authors of course. Can you list those? Collecting suggestions like this would be a start to smoothing the rough edges. Perhaps we need a common place for users' UI suggestions, where current wording is awkward or wrong, and where help improvements are needed. But as was said above, probably the best place is in the relevant thread. I think most authors welcome such suggestions.

Wow! Nice to see this, and yes it brings back memories for me too. I think Joe had the original idea, then I fleshed it out with most of them, and added the legend. He and I both tweaked it a little, but then he drastically overhauled it, in a way that was incompatible with my previous work. I was just completing a new round of improvements, but hadn't given them to him yet. Because it would take quite a bit of work to convert mine, and I didn't have time then, it was essentially abandoned, no further development. I believe he did tweak it a little later, for a few 'error' false positives. I had always wanted to get back to it, with so many changes needed, but forgot about it. I'm really glad to see you have picked it up. I hope you won't mind some suggestions, when I have more time than now. A few initial suggestions - * I strongly recommend defaulting 'emhttp' to off. Those lines are usually not important for highlighting. * Just my personal preference, but I found the initial color choices to be generally too bright or dark. That 'CSS Color Groups' page seems to have picked colors that are generally better for foreground use, not background (in my opinion). That Magenta was so bright, that after looking at a line with it, any other line had a light cyan/grey background! So here's my selections, feel free to use or ignore! But I believe they're more gentle on the eyes. Error -> HotPink (surprisingly! it's the only subdued light red choice available, but it actually works!) Minor Issues -> LemonChiffon (the only one I didn't change (sorry), good choice) Lime Tech -> LimeGreen Array -> HoneyDew System -> LightGreen File System -> Gainsboro Devices -> AliceBlue Network -> Thistle Login -> Lavender emhttp -> Azure (but I prefer disabling it completely) Other -> Seashell * I noticed you have 'Devices' instead of 'Drive related'. To me, Devices could be almost anything, and we're only talking about storage devices here. I do prefer 'Drive related', unless something better shows up. * "Tainted" is a VERY BAD thing, as far as I know. It's my understanding that it indicates a modification to the code of a module, so yes a user modified module might show up as 'Tainted' but be acceptable. But what we find here is that a module that is found to be 'Tainted' has been overwritten, a VERY BAD THING indeed! It's usually followed by a Call Trace. I've seen cases where a kernel OOPS initially reported a module was not tainted, but as the issue continued, WAS tainted, indicating serious memory or code space corruption. If I see a 'Tainted' message, I always recommend grabbing any diagnostics possible, then rebooting. It's never wise to continue, once memory or program corruption is detected. But I'm not a Linux expert, am willing to learn from others with better Linux understanding. * There's quite a bit more work needed, on the matching strings, and their order. When I have more time, I'll offer some more ideas.

squirrellydw, there have been a number of users that have posted a perfect result then asked if it's OK, and I'm never sure how to answer. There's only success messages above, nothing at all wrong, but perhaps there's too much technical info, and it's creating confusion or ambiguity? Can you tell me what caused doubt in your mind? What would you prefer it to say that would be clearer? I think we should probably push more users to use the plugin, which I believe completes with a clear and unambiguous result.

In the process of converting from RFS to XFS I also 'unjumpered' my older 2 TB WD drives and changed the alignment to 4K aligned. All is humming along That's good to hear, good to know!

from example in reply #3 let's just say if disk 4 not aligned with 4K when I finish round 4 and do round 5 "4 - Format the unformatted with unRAID. XFS disk formats." since the default MBR is 4k-aligned in settings will it format XFS along with MBR 4K ? or it will still use previous MBR ? Alignment is only relevant at drive initialization, has nothing at all to do with formatting and file systems. Alignment is only done at initial setup, with the initial GPT and/or MBR and the partitions. Once partitions are created, then you can format a partition with a particular file system. Currently, questions about alignment only come up with a previously jumpered drive. The old advice used to be - if it's not jumpered, don't add one; if it's jumpered, don't remove it, because some drives never worked correctly after the jumper was removed. Recently, some have removed the jumper, but I'm not sure how well it is working for them. It might be interesting for someone to try it the different ways, and test the disk speed (perhaps using the diskspeed script), see if it makes much difference, and if it does, which way is best.

Both drives look great, ignore the Command Timeouts, the Raw read error rates are better than good, 111 and 110 (initial perfect value is 100).

Thank you for that info, it's the first I had heard of it.

The drive claims to be usable, but I wouldn't trust it, with over 35000 remapped sectors and other indications of a serious malfunction in the past. A significant part of the drive has failed, but it's trying to work around it. The only thing I would use it for is as a third or fourth backup, or for unimportant things you don't care about losing, but not part of an unRAID array. It's not possible (outside of the factory) for a drive to change model and serial. I suspect you used the same drive symbol (sde), but that can change between boots. Either you have a 4TB drive elsewhere in your server, or the smartctl command returned garbage.

I've added a FAQ index to the first post of the FAQ thread. Unfortunately, only moderators can edit or add to it, so moderators ... please do.

All I could go by is the info you provided, the Preclear output. According to it, the old Preclear script did not recognize the drive correctly. I still think there's a better chance for that old script if the drive is on a standard non-SAS SATA port.

The drive isn't being recognized correctly, on that SAS card. Try connecting it to a motherboard port, and redo the Preclear.

It does work for most, so there must be something unusual about your board. You can help him by providing the contents of that sensors.conf, and do a sensors-detect and provide its recommendations. Try to download the latest sensors detect script.

We STRONGLY RECOMMEND upgrading to the latest v6! * ALL versions prior to the current v6 have known security vulnerabilities, making them vulnerable to attack by an infected machine on your local network. * LimeTech has been seriously considering for some time moving ALL older versions to End-Of-Life, all versions prior to v6. * Maintenance and support is becoming harder and scarcer as time goes by. There are fewer and fewer people familiar with your version able or willing to help. * Numerous bugs have been fixed. You may be running without issue now, but something can still go wrong, or you may try something different, and run into an issue long fixed in newer versions. * You're missing out on a lot of new features, and perhaps better performance too. * One very important improvement is Notifications, makes it a better NAS! The system will inform you immediately of problems. * After upgrading, you will still be able to run all of your favorite applications, and they may even run better, because in v6 there are multiple ways to run them. Please check out the Upgrading to UnRAID v6 guide.

I completely agree. A small thing, but so convenient, and helpful.

I don't remember the whole story, but there was considerable discussion about many ways to handle this, and this is what was decided. I believe it had something to do with the fact that the number of array drives used to be the primary quantity, but now you can have multiple cache drives, other managed drives, and shortly multiple parity drives, all due to LimeTech development. It's not a limitation, because Tom increased the numbers. Before you could have 6 array drives, now you can have 12 of any kind of drive. It's a change that recognizes the advances in unRAID.

Howdy Squid! When you have time, could you check my post here? It looks to me as if something is doing Docker related tasks, even though Docker is disabled, and for completeness thought I should check with you in case Community Applications is doing anything at all at array start, that might be responsible for the errors I've seen as well as the docker.img writing (with Dockers disabled). If it helps, the function responsible for the error messages is using the 'logger' tag for logging its messages.

Would you have time to create a Defect Report, outlining what you are seeing and the exact content of the messages, and include your diagnostics, so the settings/variables can be confirmed/checked?

In other words, you want to selectively move 400GB to one drive, and the rest to the other. I always prefer commander-like tools for selective copying/moving, tools like Total Commander, or the built-in MC, or the Krusader Docker. Like Total Commander, Krusader should have copying with verification.

I had a chance to try this, installed and started and tested it, and must say it's very well implemented! Nice job! In my view, there's one flaw, but perhaps it's unimportant to others, the fact that the original timestamps are not preserved. From the days of DOS and Windows, they have always been important for me, and I depend on them for certain (but not all) files, and run software that depends on their accuracy for its files. I don't know if it's possible to save them, at least the 'last date modified'? It would be incompatible with your aging feature, a nice feature too. If I'm the only one interested, then don't waste any time on it, as I may not use this after all. I have many flaws, but strangely this is not one of them, and it's extremely rare for me to wrongly delete a file (I'm sure I must have though). Plus, I'm the only user on my server. I've added it to the Upgrading to UnRAID v6 guide (Other Plugins section), and would appreciate your review of the entry. If you have any interest in adding an advanced feature, you might want to look at this topic. It might be possible to add crude auditing capabilities. Speaking somewhat ignorantly, I wonder if it's as simple as adding the option to add specific SMB options, enable logging, then provide a box for viewing the log, so that a user could look for 'who deleted what' or other info. Might be a nice feature, if easy to add. Might help catch which family member needs 'reeducating' or which employee is responsible for vandalism.

You should not have had to edit anything, and you probably don't want the syslinux from a v5 folder. You want the v6 syslinux folder with the v6 syslinux files and the v6 make_bootable. I suspect the Upgrading to UnRAID v6 guide may work better for you. Check the Files on flash drive section, to make sure the right files are in the right place.