August 12, 200817 yr Author > If the drive temp is over 40C, it is highlighted in yellow, if over 50 in orange, and if over 60 in red. I think 60c is too hot to be in red, This is a dangerous position to be in, On many of my scsi drives they sihutdown on their own at 55c. Is there any way the warning(color) values could be loaded from a config.
August 12, 200817 yr > If the drive temp is over 40C, it is highlighted in yellow, if over 50 in orange, and if over 60 in red. I think 60c is too hot to be in red, This is a dangerous position to be in, On many of my scsi drives they sihutdown on their own at 55c. Is there any way the warning(color) values could be loaded from a config. It was late when I set the initial ranges... I agree, 60C is too high for the upper temp. The actual function to get the drive temp is below. You can see I've changed the temp settings for warnings already to yellow=40C, orange=45C, red=50C. You can adjust them as you like once you have the program on our own server. Of course, now that you can see part of the program, you will find it easier to figure out the language I used to write the new web-server. Joe L. [b]function GetDiskTemperature(theDisk, the_temp) { # Warning temperature ranges are set here yellow_temp = 40 orange_temp = 45 red_temp = 50 the_temp="*" is_sleeping = "n" cmd = "hdparm -C " theDisk while ((cmd | getline a) > 0 ) { if ( a ~ "standby" ) { is_sleeping = "y" } } close(cmd); if ( is_sleeping == "n" ) { cmd = "smartctl -d ata -A " theDisk "| grep -i temperature" while ((cmd | getline a) > 0 ) { delete d; split(a,d," ") the_temp = d[10] "°C" if ( d[10] > yellow_temp && d[10] <= orange_temp ) { the_temp = "<div style=\"background-color:yellow;margin-right:8px;\">" the_temp "</div>" } if ( d[10] > orange_temp && d[10] <= red_temp ) { the_temp = "<div style=\"background-color:orange;margin-right:8px;\">" the_temp "</div>" } if ( d[10] > red_temp ) { the_temp = "<div style=\"background-color:red;margin-right:8px;\">" the_temp "</div>" } } close(cmd); } return the_temp } [/b]
August 12, 200817 yr Would it be possible to show the size and possibly time/date fileds of the dupe files so we can avoid tracking down "different" files with the same name? My music collection is up to near 4tb's now and there are a LOT of folder.jpg files!!! Better yet, how about not even showing them if they have different sizes... It appears as if unRAID only puts an entry in the syslog for the subsequent file(s) with a duplicate name, not for the first. Let me think about this. It might be VERY slow to look up thousands of file sizes since I'll have to issue an "ls -l" command for each. Joe L.
August 12, 200817 yr Next idea: Show the disk spin down settings for the drives (unRAID not turned on this now so cant check). The official unRAID has a scattergun approach of settings all drives the same but I can see scenarios where it would be better to have finer granularity. Stage one would be visibility of the per drive settings. I do not know if the current spin-down time for any specific drive is available anywhere. The hdparm command sets it, but nothing I know of reads it. Also have you considered a cron job to precache alot of this info. What I am thinking is that with a cron job logging stuff you start to get some cool stats like: How long since last drives was spun down, how much data has been transferred in the last hour, whats the temp increase of this drive. Once we have the data the data manipulation and presentation options are limitless. No, had not considered collecting statistics. That is a different project than the real-time display of the statistics, and easier once you see examples of how to collect it. You would just not need to wrap the data with HTML, and no dealing with the TCP/IP networking. Joe L.
August 12, 200817 yr Author Of course, now that you can see part of the program, you will find it easier to figure out the language I used to write the new web-server. LOL, I knew that if it wasn't a shell script, it had to be gawk due to the complexity of what you are writing. What I learned (from you) was that it can do network I/O. (which is pretty slick!).
August 12, 200817 yr Of course, now that you can see part of the program, you will find it easier to figure out the language I used to write the new web-server. LOL, I knew that if it wasn't a shell script, it had to be gawk due to the complexity of what you are writing. What I learned (from you) was that it can do network I/O. (which is pretty slick!). I spent a lot time the other night doing another google search looking for other examples of network code using awk. I found quite a few instances of the code examples as described in the "white-paper" describing the TCP/IP capabilities in awk, but after tracing down a few hundred matches in google I found only three or four people who have done anything substantial in it other than those examples. It is apparently a well guarded secret, and since there are so few examples, it goes unused. I guess "perl" is the preferred language for this type of task, but then, it is a large package to add to an unRAID server, and awk is included. Besides, I never learned "perl" (although I do remember sending an e-mail to Larry Wall about it long ago when he first wrote about it in USENET newsgroups. It did not do networking back then either) Joe L.
August 12, 200817 yr Author I guess "perl" is the preferred language for this type of task, but then, it is a large package to add to an unRAID serve It would be "preferred", but then again, it's a larger package to install and gawk is already there on unRAID. I would suggest you look at perl closer. Perl is like sh/sed/awk all in one with major extensibility. I've converted awk cron scripts to perl daemons and saved hundreds of CPU cycles in the process. Once you get the hang of it, you will love it.
August 12, 200817 yr This is off-topic, so apologies are needed here, but I felt a comment about dupe files was needed. I have something like 10,000 duplicate files so syslog is exploding. I think there may be a misconception about dupe files listed in the syslog. Dupe files that show up in a syslog are NOT acceptable, and need to be removed. A dupe file here is simply a file with the exact same file name as another file in the exact same folder of the User Share. That has NEVER been allowed in any file system I am aware of. A differing file size or file contents or physical location is irrelevant. When unRAID is creating this virtual file system for the User Share, the folders must still obey standard file system rules. Tom is simply logging the subsequent file entries that match an existing file in the virtual folders, created for the User Share. If you have numerous duplicates, then you likely have multiple copies of identical or similar folders, and all that has to be done is rename the top level folder for a duplicated tree, to something like BACKUP OF PICTURES, etc. Just changing the path name in the slightest way, will remove the dupe files, and still allow you to have backup copies. By the way, having thousands of dupes is a dangerous, unstable situation. Just stopping and starting the array will explode the syslog, possibly using up all available memory, and crashing the server. I would deal with this as soon as possible. Definitely do not Start the array without rebooting, starting a fresh syslog.
August 12, 200817 yr yup your right. i know where they came from but its not so trivial to parse them out as just a rename. Essentially I am on it but it will take time. I dont reckon many people will be like me but im sure lots of people will have a few so in that respect it fits in well with general house keeping. I have 4GB of memory so it would be virtually impossible for syslog to fill it up but I hear what your saying and its well up on my to do list. But this now temperture is a more critical problem for me and thats what im am experimenting with every day.
August 12, 200817 yr The issue is not that you have two files with the same name, but two files name the same in different /mnt/disk[123456789]/Share/ folders Example: You have a share named "Music" and it exists on /mnt/disk1/Music and also on /mnt/disk4/Music. When unRAID creates a "user share" it merges the contents of the two like named "disk" folders onto /mnt/user/Music. If /mnt/disk1/Music/Oldies has a file named "favorite.mp3" and /mnt/disk4/Music/Oldies also has a file named "favorite.mp3" only one can be made visible in the user share as /mnt/user/Music/Oldies/favorite.mp3. The second file is the one reported in the syslog as a duplicate. It might be the same song, it might not. There is no way to tell except by playing them, looking at their size, checksum, etc. To fix this, all you need do is rename one of the two disk folders to "Oldies2" To do this, type: mv /mnt/disk4/Music/Oldies /mnt/disk4/Music/Oldies2 Now, in user shares you will see "/mnt/user/Music/Oldies/favorite.mp3" and also "/mnt/user/Music/Oldies2/favorite.mp3" You might have identical named top level folders with identical backups of precious files in /mnt/disk1 and /mnt/disk2. To fix the 10000 entries user shares is complaining about as duplicates might only need a rename of one or two upper level folders. Joe L.
August 12, 200817 yr Author I wonder if it's worthwhile to virtually rename the duplicate files to something like name~1 name~2, etc, etc... This way you can get to the file via share and remove the duplicates easily.
August 12, 200817 yr OK, attached here is a "beta" version of my awk based web-server (actually gawk) to play with... It grew to 728 lines of code, but that includes a few comments. For the trivia fiends out there, "awk" was named by taking the last initial of the three authors who wrote it back at Bell Labs. Alfred V. Aho Peter J. Weinberger Brian W. Kernighan Although Al Aho is not too well known (he was the original author of "egrep" and "fgrep"), Pete Weinburger originally wrote libPW. (the regular expression library) Most will know Kernighan, he wrote the book on "C" language... I think he did a tiny bit of UNIX as well. To understand it, all you need to know is a bit of logic, a bit of networking, a bit of "css", a tiny smattering of HTML. It is readable otherwise. It is fairly modular and not too hard to add additional pages. The disk management menu page is partially in place. It does not manage unassigned disks, it does not yet have the ability to unmount, check, or repair file systems... but it will some day. Everything else should work. If it does not, let me know what it does/doesn't do. As I said, it will probably show the cache drive as unassigned... (nobody got back to me with how the disk.cfg file looks when you have a cache drive defined... YET. (please)) If you extend it and add features, please share back with the unRAID community. Consider it to be GPL, much like the "md" driver. # It is not guaranteed to be able to do anything, # but it does not have a "Restore" button, therefore, it is an improvement # invoke as: awk -W re-expand -f unmenu.awk or awk -W re-expand -v MyPort=90 -f unmenu.awk # point browser to //tower:8080 or //tower:90 (if you designated your own port on the command line) newest version is attached here: http://lime-technology.com/forum/index.php?topic=2595.msg20982#new Joe L. Disk Management Menu
August 12, 200817 yr I wonder if it's worthwhile to virtually rename the duplicate files to something like name~1 name~2, etc, etc... This way you can get to the file via share and remove the duplicates easily. I want to fix this by removing the duplicate data and since all data under the duplicate folder name is not all duplicate renaming will solve the syslog issue but I really dont want to leave 300GB+ of dupe data kicking about. Anything that helps solve that is a boost but i cant see many people letting it get as bad as I have so effort vs. need should be taken into account. I am going to manaully parse out and fix this in the next 24 hours. Joe I had real doubts how far you could take this with the method you chose to deliver the pages. I am glad to say I couldn't have been more wrong. Superb work; and i don't say that lightly. WeeboTech and i discussed briefly releasing a tarball of the new folder structure so that users had a simple means of having it on their keys. Can i suggest that since this new GUI is likely to be the most popular add on ever we combine the two and kill two birds with one stone for later releases of your tool?
August 12, 200817 yr Joe I had real doubts how far you could take this with the method you chose to deliver the pages. I am glad to say I couldn't have been more wrong. Superb work; and i don't say that lightly. It actually looks pretty simple, doesn't it... at least for all it does. Trust me, nobody will ever think it is an "awk" script serving up web-pages. WeeboTech and i discussed briefly releasing a tarball of the new folder structure so that users had a simple means of having it on their keys. Can i suggest that since this new GUI is likely to be the most popular add on ever we combine the two and kill two birds with one stone for later releases of your tool? Sounds fine, I'm sure it will evolve a bit at first, as people think of what might be added. Goal is to supplement emhttp, and not to do anything to make Tom worry. We'll have to figure out how to invoke it without a controlling terminal. Probably "nohup ..... &" will do it. Not sure if I mentioned it, but if an "unassigned" disk is present, the filesystem type on it is displayed. Did not need to do that with the "md" disks as they all have a reiserfs. A long time ago my old manager at AT&T bet me I could not write a "btree" based relational database using unix shell (and the built in commands on unix) He lost the bet. (You can use "dd" to seek and read/write bytes wherever you want) Joe L.
August 13, 200817 yr Joe - Just had to say again that this looks truly outstanding. I think you made a smart choice to use a feature of a component in the delivered solution, rather than requiring the installation of another component. This will likely increase "usership" substantially. I don't know awk very well, but look forward to learning a bit in order to make my desired tweaks. On that topic, you might want to give some thought as to how the solution could be made modular and extensible. You know that a lot of people are going to have a lot of requests for a lot of new features. Some people are going to go hacking on the script to add the features that they want. Sharing useful pieces is going to be tricky is this "wild west" model. Not sure if there is some way to provide "exits" and/or standards to make sharing of useful features / pages easier. Are you planning on collecting the pieces and producing "unified" versions over time? (No pressure, just wondering what you were thinking). Thanks again for all your hard work. Looking forward to giving it a try. - Brian
August 14, 200817 yr I did a bit more work on the "awk" (gawk) based web-server for unRAID. I now have more of the Disk Management page functional. It still only deals with devices assigned to the unRAID array. Unassigned devices and the cache device can not be managed from that page at this time. I removed the "Un-Mount" button. It would only fail if SAMBA was still enabled. I figured the only time you would want to un-mount a drive is when you were performing a file-system check (or repair) on it, so I automated the process. So... the "File-System-Check" and "File-System-Repair" buttons now will: Stop Samba Kills any process accessing /mnt/user or the disk being tested. (This ensures we can un-mount the drive) Un-Mounts the drive Runs reiserfsck on the /dev/md? device. This keeps parity up to date even if a drive is repaired by the file-system check. Re-mounts the drive Re-Starts Samba. While SAMBA is stopped, no User or Disk shares will be visible on the LAN. The process of performing a file system check or repair can take from several minutes to an hour or more. During this time, the browser will appear to be "waiting" Be patient, it will eventually return with the results. (it takes about 35 minutes to do a File-System-Check on one of my 500G drives that is nearly full) The results are logged to /var/log/syslog too. Do not run a file system check on the parity drive... it has no file-system. ( It will try to unmount it, but since it is not mounted, it can't, it will try to re-mount it, but since it had no "md" device, that will fail too... In the end, the parity drive will be unaffected. It will stop and re-start SAMBA while it tries to work on the parity drive, so be aware. It does verify that the first disk partition has a reiserfs before running reiserfsck on it, so you should be safe) If you access the normal unRAID management page while a disk is unmounted and being tested it will show it as "unformatted" Do not press the button to "format" it. Also, Tom's normal unRAID management page has no indicator showing SAMBA has been stopped. If you stop SAMBA using this utility, the only way to re-start it using the normal management page is to Stop the array, and then Start the array. Do not Stop the array or Start the array using Tom's web-management utility while in the middle of doing a File-System-Check. I really don't know how badly confused it will get, but I have not tested every possible outcome... Let the file-system checks complete, even if you have to wait a while. If something goes wrong, kill this awk based web-server, Stop the array using Tom's normal web-management page, and then re-start it using the normal web-management page. That will re-mount the drives. I take no responsibility for what damage you can do with this tool. It unmounts a drive and runs reiserfsck. Neither should have any effect on your data. You should not run the "File System Repair" on a disk unless you first run a "File System Check" and it advises a repair is needed. The "File System Repair" button is equivalent to running "reiserfsck --fix-fixable -q -y /dev/md? 2>&1| tee -a /var/log/syslog" on a drive. I'll attach a new version of the unmenu.zip file with these changes after I get some feedback or comments on my changes from those following this thread. (This way, I can react to any good ideas.) Joe L.
August 14, 200817 yr So far so good here. Works really well and is easy to launch. Couple of new ideas (usual caveat of might be silly/flights of fancy apply here): Being able to run the ls -R trick via the gui. Would need consideration on what folders to run it on though so unless you can load some sort of config this might not fly. Stats on user shares. For each user share show interesting stats like: size on each disk, allocation method, total size, numbers of files etc. Might not be viable without caching Summary of log file sizes since we know certain things can explode them. Summarised SMART stats. rather than show a SMART stats per drive have the option to see all of field xx for all drives in one go. Interesting stuff like uptime, kernel version, CPU, memory installed etc
August 14, 200817 yr Since no major feedback about bugs, attached here is the newest version of the awk based web-server for unRAID. I've cleaned up the code in a few spots, and probably made it more confusing in a few others. In all, it should be readable to fellow geeks. Have fun.. invoke as before, All functions present should work. Let me know if otherwise. It does NOT configure unRAID, but reports on it, and allows some additional management tasks not present on the stock web-management utility. newest version is attached here:http://lime-technology.com/forum/index.php?topic=2595.msg20982#new Joe L.
August 14, 200817 yr Thanks so much for your efforts on this, Joe & for your generosity in sharing it. Can't wait to check it out when I get home. Best bb
August 15, 200817 yr Some comments and suggestions: Very good job! It immediately inspires more ideas. I'm very tempted to add more to it, but I'll send anything directly to you for your control. Would it not be wise to add version info to your initial comment section? Anything of a non-trivial nature, with ongoing editing and releases, should really have some sort of version control. Since this is the second generation of unmenu, I've temporarily assigned 2.10 to my copy of the August 12 release, and 2.11 to today's release, but will change to whatever you prefer. You might consider making the unmenu version visible to the user too, perhaps close to the display of the unRAID and Linux kernel version, after you add them! Perhaps an About page? What do you use or what would you suggest for an awk reference manual/man page/etc? In function SetUpDiskUsagePage(), should PageDocc be PageDoc? Rather than tell users not to perform invalid actions (such as "File-System-Check" of the parity drive), would it not be safer to avoid displaying any invalid options. It would be good to detect the parity drive, and only show a very limited set of options. We used to call this 'idiot-proofing'. Concerning your "File-System-Repair" functionality, my experience with reiserfsck is considerably less than yours, but I have seen cases where reiserfsck has advised use of the --rebuild-tree option, as well as cases where --fix-fixable was advised. Since you are only including the --fix-fixable option, I'm assuming that is the more likely case, but what if --rebuild-tree is what it wants? I'm a little uncomfortable with either choice being too easily selected by new users though, especially if "File-System-Check" has not been run, and even if reiserfsck has run, but had no further directions. And are there possibly other rare reiserfsck options that might be advised also? Perhaps add somewhere (initial splash screen?) a strong warning about not using management functions of both unmenu and the unRAID web management pages at the same time. An event loop type program would be nice here, to monitor array status and watch for changes made from the unRAID web page. Can awk provide anything in the background here, to 'trap' status changes? (Very rusty, I'm using terminology very loosely here.) It would be nice to see the array status (Stopped/Started or similar) in the header of every page, and dynamically change the choices to match the status, safer too. We don't want to make it easy to create more tech support problems, by a careless or impulsive new user, who likes to push buttons to see what they do. I do realize it will be very difficult to idiot-proof unmenu.
August 15, 200817 yr Joe L. - This is great! Very easy to install. Excellent functionality. I do have a few issues to report. 1. On the main page, the drive sizes all seem to report a tiny bit OVER the actual size. 300G reports as 301G, 500G as 501G, 750G as 751G, and 1T as 1.1T. 2. On the main page, the total line is blank. 3. When I cliicked the "Dupe Files" tab, it never went to the dup files page. I thought it was probably working on it, but looked at Tom's emhttp page and didn't see any read activity. I killed the process and tried again - same result. Odd but later on I clicked the dupe files page by accident, and the page came up. (There are no dup files in my syslog - so if that's all it is looking for - it should be real quick). 4. When I cliicked the "Disk Usage" tab, it seems to start reading the drives hard one at a time. I have not gotten it to finish yet. Not sure what this is doing, but will try and let it finish so that I can see it. If I click on another page seems I have to kill the process and run it again. 5. The Web pages never seem to fully resolve - the windows "flag" in the upper right never seems to stop. Perhaps this is by design. Hope this helps. If you need a screenshot of my main page, let me know.
August 15, 200817 yr Some comments and suggestions: Very good job! It immediately inspires more ideas. Exactly my goal. Obviously, I can manage my array from the command line; I did not need a web-based tool. I certainly did not need to figure out how to do it with awk. This exercise is to give everyone, including Tom, a few ideas of how the web-interface might evolve and to do it in a way that does not involve emhtp so we do not need to take time from his own work. I'm very tempted to add more to it, but I'll send anything directly to you for your control. Please, make it better, faster, nicer. It is actually pretty easy to add a new"page," especially if it does not contain any buttons. It is slightly more complicated if it has a form and input fields. Would it not be wise to add version info to your initial comment section? Anything of a non-trivial nature, with ongoing editing and releases, should really have some sort of version control. Since this is the second generation of unmenu, I've temporarily assigned 2.10 to my copy of the August 12 release, and 2.11 to today's release, but will change to whatever you prefer. You might consider making the unmenu version visible to the user too, perhaps close to the display of the unRAID and Linux kernel version, after you add them! Perhaps an About page? An about page sounds great. Do you want to try adding it? It is not too hard. One function for the html content, a clone of an existing menu line for the top menu, and an additional "if" statement to invoke the new page. What do you use or what would you suggest for an awk reference manual/man page/etc? http://www.gnu.org/software/gawk/manual/gawk.html In function SetUpDiskUsagePage(), should PageDocc be PageDoc? Oops... yes, you are correct. Rather than tell users not to perform invalid actions (such as "File-System-Check" of the parity drive), would it not be safer to avoid displaying any invalid options. It would be good to detect the parity drive, and only show a very limited set of options. We used to call this 'idiot-proofing'. I know the concept, having had to deal with developing software applications for corporate intranet use for over 20 years. I just did not add any Javascript logic, yet. Concerning your "File-System-Repair" functionality, my experience with reiserfsck is considerably less than yours, but I have seen cases where reiserfsck has advised use of the --rebuild-tree option, as well as cases where --fix-fixable was advised. Since you are only including the --fix-fixable option, I'm assuming that is the more likely case, but what if --rebuild-tree is what it wants? I'm a little uncomfortable with either choice being too easily selected by new users though, especially if "File-System-Check" has not been run, and even if reiserfsck has run, but had no further directions. And are there possibly other rare reiserfsck options that might be advised also? I felt the --fix-fixable was fairly safe, the rebuild-tree is much more drastic. In reality, neither should cause problems if run on a good file system, but I really did not want to put the rebuild-tree choice on the web-interface. Perhaps put the file system repair on its own page Perhaps add somewhere (initial splash screen?) a strong warning about not using management functions of both unmenu and the unRAID web management pages at the same time. That might work. An event loop type program would be nice here, to monitor array status and watch for changes made from the unRAID web page. Can awk provide anything in the background here, to 'trap' status changes? (Very rusty, I'm using terminology very loosely here.) It would be nice to see the array status (Stopped/Started or similar) in the header of every page, and dynamically change the choices to match the status, safer too. We don't want to make it easy to create more tech support problems, by a careless or impulsive new user, who likes to push buttons to see what they do. I do realize it will be very difficult to idiot-proof unmenu. I do update the status in the top of the screen on each refresh,and I do update and display only the appropriate choices for any given status already. I did not yet add an auto-refresh, but I think I will, as long as the array status is "STARTED" I think it is safe. I know it is not safe to check status while the array is expanding a drive. The issue is not with my awk based screen, but in the emhttp screen interpreting any unmounted file system as "unformatted". It is on it where a user could "format" a drive. Joe L.
August 15, 200817 yr Joe L. - This is great! Very easy to install. Excellent functionality. I do have a few issues to report. 1. On the main page, the drive sizes all seem to report a tiny bit OVER the actual size. 300G reports as 301G, 500G as 501G, 750G as 751G, and 1T as 1.1T. Log onto linux, type df -H What you see is what I report. All I can figure is that it rounds up. I don't love it, but I like it better than df -h 2. On the main page, the total line is blank. Do you have user shares enabled? On my server, the df -H command has a line for "shfs" on /mnt/user. That is what I show on the total line. 3. When I cliicked the "Dupe Files" tab, it never went to the dup files page. I thought it was probably working on it, but looked at Tom's emhttp page and didn't see any read activity. I killed the process and tried again - same result. Odd but later on I clicked the dupe files page by accident, and the page came up. (There are no dup files in my syslog - so if that's all it is looking for - it should be real quick). it just looks for lines in the syslog matching "duplicate object" 4. When I cliicked the "Disk Usage" tab, it seems to start reading the drives hard one at a time. I have not gotten it to finish yet. Not sure what this is doing, but will try and let it finish so that I can see it. If I click on another page seems I have to kill the process and run it again. It is doing: du -s --si /mnt/user/* /mnt/disk*/* 2>/dev/null | awk ' { print $2 \"|\" $1 }' | sort It will take a while. It is reading the size of every file on every disk... (probably twice, once on the /mnt/disk share, the other on /mnt/user. 5. The Web pages never seem to fully resolve - the windows "flag" in the upper right never seems to stop. Perhaps this is by design. Never noticed that on my browser. What browser are you using? Don't have a lot of control over it... Hope this helps. If you need a screenshot of my main page, let me know. Don't need a screen shot, but a "df -H" output would be appreciated. Joe L.
August 15, 200817 yr 1. Yes, "df -H" does indeed report the values you are showing on the main page. The +1g isn't so bad (e.g., 500G -> 501G), but the +100g (1T -> 1.1T) is pretty seriously wrong. Although I'd like the rated size to show, I think that the "df -h" results are more accurate. 2. I do not have user shares enables, I use the disk shares. Is there a way to show the total size of the array without relying on user shares? 3. 10-4. The issue is that the screen doesn't show sometimes. If this is just me, no problem (with no users shares, I have no duplicates in my syslog). 4. Disk usage took a while but worked. It is a very nice feature! Given that this takes a while, from a UI perspective, it might be nice if the disk usage screen were presented and had a button to do the data collection. The screen could explain that it takes a while. This is a nit. 5. I am using IE6. I also sent you a PM. Brian
August 15, 200817 yr 1. Yes, "df -H" does indeed report the values you are showing on the main page. The +1g isn't so bad (e.g., 500G -> 501G), but the +100g (1T -> 1.1T) is pretty seriously wrong. Although I'd like the rated size to show, I think that the "df -h" results are more accurate. Whoever wrote the routine to convert the values to human readable ones must have had a marketing background It is rounding up, even after you calculate block sizes in powers of 10 rather than powers of 2. You can see it here: df --block-size=10000000 /dev/md6 40008 39021 987 98% /mnt/disk6 df --block-size=100000000 /dev/md6 4001 3903 99 98% /mnt/disk6 df --block-size=1000000000 /dev/md6 401 391 10 98% /mnt/disk6 2. I do not have user shares enables, I use the disk shares. Is there a way to show the total size of the array without relying on user shares? I'll probably have to add the column totals myself to do that. Might give it a try, might even be more accurate in how I round the numbers. 3. 10-4. The issue is that the screen doesn't show sometimes. If this is just me, no problem (with no users shares, I have no duplicates in my syslog). Not sure what is happening. 4. Disk usage took a while but worked. It is a very nice feature! Given that this takes a while, from a UI perspective, it might be nice if the disk usage screen were presented and had a button to do the data collection. The screen could explain that it takes a while. This is a nit. I like that idea a lot. Look for it in the next version. The basic idea for it was not mine, it was NAS. He suggested: Stats on user shares. For each user share show interesting stats like: size on each disk 5. I am using IE6. I also sent you a PM. Brian Got the PM, good idea about moving this to User-Customizations. I'll do it later, once I get a few more tiny things done, for now, it ties very well to the title of this thread for "Revamping the initial unRaid page" ... It is just that we are doing it rather than Tom.
Archived
This topic is now archived and is closed to further replies.