OK, here goes...  I've named this version 1.0beta.   Lots of loose ends are now completed and in place.

 

I've had to make changes in several of the plug-ins, so I've included them all once more.   

 

New in this release is the ability to manage disks that are not in the protected array.     If a disk is partitioned and has a file-system on it, and the file-system is one we know how to deal with, we can mount it, and then create a SAMBA share for it.  We can then un-share it, and then un-mount it.  This will greatly help those who have external USB drives with older data  from other PCs.

 

Note: the file-systems are mounted as "read-only" by default, but you can add an entry to the unmenu.conf to mount them as read/write.

ReadOnlyMount = -w

To mount an NTFS partition will require the installation of the ntfs-3g package. (Not sure if Tom is supplying that in his base release yet..  If not, it is easy to download, and one command will install it)

 

It will name the mount point after the disk partition.  If, for example, the disk partition is /dev/sdd1, it will be mounted on /mnt/disk/sdd1

 

The share it creates will then be known as "sdd1" on your LAN, it is also named after the disk partition.

 

If the drive has the ability to run SMART or HDParm commands, buttons for those functions will also be available on the base partition of the drive...  You can now install a new SATA or IDE drive in your array, run SMART tests on it, and then get the result using the "Smart-Status" button.   (these buttons are not present on the USB and SCSI drives, since to my knowledge, SMART functions are not supported.)

 

When downloading  "/var/log/syslog", it is now given a name based on the current date.  The "registered to:" line is removed for added privacy.

 

Once I get some feedback that everything seems to work as expected,  I will create a new thread in the "User Customizations" forum so this has a bit more visibility.  As it is now, it is buried in this thread with tons of suggestions for improving the unRaid management page, most of which have been implemented in some way. 

 

I'm open to ideas on how the management of "un-assigned" disks can be improved.  I'm sure we can add some file-system-checks, for those where we have the software.  I do not have any provision for clearing, partitioning, or formatting drives.  (I'll leave that to Tom and emhttp)

 

Have Fun...  feedback is encouraged.  This is a project where everybody can contribute.

 

To run this:

un-zip and put all the files in a folder on your USB flash drive..  (I just have them all at /boot for now)

log-in via telnet and change directory to that folder  cd /boot

Type the following command to start the unmenu.awk web-server

awk  -W re-interval -f unmenu.awk

 

Using your browser on your PC, enter the following URL

http://tower:8080/

 

 

Attached is the latest version, The main page now looks like this:

 

The disk management page looks like this. The buttons presented are based on the disk and each partitions status.  In this example, I pressed the "Mount" button, and now I have options to un-mount, or to create a SAMBA share.  If I were to create a share, I would only have an option to un-share it.  If not mounted, you only have an option to mount it, and only if it has a file-system type recognized.  Hopefully, I got the logic right for when to present the various buttons.

 

Newer Version, with fixes in this post: http://lime-technology.com/forum/index.php?topic=2595.msg20982#new

Joe L.

Lookin' mighty spiffy there Joe!!! Awesome work!

I don't understand the total disk count. It says I have 14 disks in the array, when I actually only have 11. I'm attaching a screenshot. Other than that, it looks awesome. 

I don't understand the total disk count. It says I have 14 disks in the array, when I actually only have 11. I'm attaching a screenshot. Other than that, it looks awesome. 

Oops... I knew somebody would find something I overlooked. 

Clearly I am counting the array slots, and not considering the unpopulated slots you left empty.

I do not have an array with the same situation, but... I'll fix that for the next release...

 

All I can say is thank you for your help in testing this unique program. 

If you find anything else, let me know, as I'd love to soon say I finally have release 1.0 finalized.

 

Joe L.

 

Edit: I fixed it in my version... I'll wait a tiny bit before posting the 1.0-final, just in case.   In the interim, mentally subtract the three unpopulated slots in the middle of your array...

Small nit. File browser page doesn't keep the links menu consistent at the top of page, only shows Main.

Also, it seems to have a problem showing files or folders with spaces in the name.

 

Also, I get this message on the system info open files plugin "you don't appear to have the "lsof" package installed"

I'm not saying this is bad behavior, quite proper in fact. Just noting that it's not installed by default, and you may want to add a quick note on how to install it, or a link to the instructions. This is the only reference I found in a quick search of the forums.

http://lime-technology.com/forum/index.php?topic=2312.msg17691#msg17691

 

Same comment in regards to the PCI Devices plugin "you don't appear to have the "lspci" package installed"

 

Thanks for your hard work on a very useful tool!

I notice that the plugins are rather sluggish.  They each take about 8 seconds to load.  I don't think the earlier versions had this lag.

 

This is not a complaint - just wondering if others are noticing something similar, and whether there might be a bug causing this slowdown.

Small nit. File browser page doesn't keep the links menu consistent at the top of page, only shows Main.

True... 

The file browser plug-in supplies the entire content for the web-page, including the HTTP header and the HTML content.  It is the only plug-in that does everything in itself.  In effect, it is its own web-server.  All of the other plug-ins supply only the HTML for the middle portion of the screen and the unmenu.awk script supplies the HTTP header and the entire top of page, including the links representing all the other plug-ins.    I did not want the file-browser to have to scan all the plug-ins, and build a top menu, as it would only slow things down as you moved from directory to directory browsing.  So, the omission of all the plug-in links on the top menu of the file-browser is basically to make it a bit more efficient (and faster).

Also, I get this message on the system info open files plugin "you don't appear to have the "lsof" package installed"

I'm not saying this is bad behavior, quite proper in fact. Just noting that it's not installed by default, and you may want to add a quick note on how to install it, or a link to the instructions. This is the only reference I found in a quick search of the forums.

http://lime-technology.com/forum/index.php?topic=2312.msg17691#msg17691

 

Same comment in regards to the PCI Devices plugin "you don't appear to have the "lspci" package installed"

Good idea for instructions...  Here are two lines you can add to the unmenu.conf file  (each should be on its own line)

[pre]

USEFUL_LINK = <a href="http://packages.slackware.it/package.php?q=current/lsof-4.78-i486-1"><b>lsof</b> install package (list of open files)</a> download, then install b

y typing  <b>installpkg lsof-4.78-i486-1.tgz</b>

USEFUL_LINK = <a href="http://packages.slackware.it/package.php?q=current/pciutils-2.2.10-i486-1"><b>lspci</b> install package (list of PCI interfaces)</a> download, the

n install by typing <b>installpkg pciutils-2.2.10-i486-1.tgz</b>

 

[/pre]

Thanks for your hard work on a very useful tool!

You are welcome.

 

Joe L.

I notice that the plugins are rather sluggish.  They each take about 8 seconds to load.  I don't think the earlier versions had this lag.

 

This is not a complaint - just wondering if others are noticing something similar, and whether there might be a bug causing this slowdown.

Earlier versions did not use the unmenu.conf file, and, even if they did, it had less content.  I suspect that is the biggest difference.

 

Basically, many of the plug-ins are now opening up the file and parsing and reading in all the configurable variables, so it is doing a lot more...    If it also happens on a plug-in that does not look at unmenu.conf, then something else might be the cause.  plug-ins are not as efficient as built-in pages.  To invoke a plug-in, the main awk program invokes a sub-shell, using sh -c "awk ..... plugin.awk" so there is a fork and exec of the shell, followed by a fork and exec of the awk plugin.

 

I'll certainly keep an eye open on making any improvements I can for efficiency, but the very fact it serves a web-page at all,  in any reasonable time is a small miracle...

 

Joe L.

Good idea for instructions... [about packages]

 

How about lm-sensors? After finding it at the same site and installing it, I still get no sensor feedback in System Info -> CPU Info. Running /usr/bin/sensors on the cmd line brings a mention of sensors-detect, which in turn seems to need perl. Is that how to do it, if you don't mind me asking, or is there an easier way?

 

(In case it's relevant, I'm using a P5B-VM DO.)

Joe L,

I haven't gotten a chance to try this, but it looks pretty cool.  I hope to give it a try this weekend.

 

Thanks

I notice that the plugins are rather sluggish.  They each take about 8 seconds to load.  I don't think the earlier versions had this lag.

 

This is not a complaint - just wondering if others are noticing something similar, and whether there might be a bug causing this slowdown.

I'm glad you said something...  I did make a change recently, in an attempt to make the program more efficient...  boy was I wrong.

 

The data from the plug-in is read (currently) in 1024 byte "chunks" by unmenu.awk.  It used to be in 512 byte chunks, and I doubled it figuring it would have to switch less frequently from the child process to the parent.  Apparently, that was the cause of the slowdown.

 

Further testing indicated I should use a smaller buffer than 512.... I'm not sure what the fastest size is, but try this simple change:

 

Around line 282 of unmenu.awk make the following change

from

                  RS = ".{1,1024}" # read 1024 bytes at a time from the plug-in, this should allow us to handle binary data

to

                  RS = ".{1,128}" # read 128 bytes at a time from the plug-in, this should allow us to handle binary data

 

It makes a HUGE difference when reading 128 bytes at a time vs 1024..  Pages load MUCH faster... within a second or so instead of 5 or more seconds.

 

Joe L.

Good idea for instructions... [about packages]

 

How about lm-sensors? After finding it at the same site and installing it, I still get no sensor feedback in System Info -> CPU Info. Running /usr/bin/sensors on the cmd line brings a mention of sensors-detect, which in turn seems to need perl. Is that how to do it, if you don't mind me asking, or is there an easier way?

 

(In case it's relevant, I'm using a P5B-VM DO.)

As I did not write the system info plug-in, I'm not the best person to answer.  I ran into the same issue as you.  I tried installing lm-sensors, and it seemed to depend on other packages, as you discovered. I don't have it working either.

 

bubbaQ might be the best person to help us here...

 

I think a package-install plug-in might be a great addition.  It would assist in the download, and installation of useful packages.

 

I suspect that unmenu.awk plug-in collection will grow over time... and plug-ins will evolve for all kinds of tasks that might be difficult for beginners otherwise.  Even if unmenu.awk itself is replaced, functions and features will live on in some replacement.  Based on what I've seen so far, the "awk" based server it is pretty capable... and with plug-ins being almost trivial to write, there will be many written for local needs.

 

Joe L.

Based on what I've seen so far, the "awk" based server it is pretty capable...

 

That it certainly is! And, very much to your credit, the added functionality is extremely useful and extremely well executed. Thank you!

just a +1 on the before-mentioned file browser problem with files/folders that have spaces in the name...

 

 

Small nit. File browser page doesn't keep the links menu consistent at the top of page, only shows Main.

I now pass the top menu links as an argument to the plug-in process so it can have the same menu.  It will now look consistent in the file browser. (in 1.0-final)

Also, it seems to have a problem showing files or folders with spaces in the name.

I do not seem to have that problem.   More details please... What browser are you using?

Can you supply a screen shot?

 

Joe L.

What browser are you using?

The one built into unmenu 

 

The screen shot would be an empty page if the folder has a space in the name... or a missing file if the file has a space in the name 

 

 

I spoke too soon. It does not work here either... but I fixed it.

 

Spaces in file names in the file_browser plug-in used to work until I added the following function and used "c" as the array name.  This clobbered a different use of the variable "c" in a different context.  Guess I just did not hit it in subsequent usage here...   Glad you are are helping me.

 

I re-named the variable as "f" below... it is the last function in the file_browser plug-in.

 

So, in 30-unmenu-file_browser.awk, replace the existing GetAllowedFolder function with the one below, or... just grab the newest version of everything in attachment in the next post....

 

 

function GetAllowedFolders(cfile) {

    RS="\n"

    while (( getline line < cfile ) > 0 ) {

          delete f;

          match( line , /^(ALLOWED_FOLDER)([\t ]*)(=)([\t ]*)(.+)/, f)

          if ( f[1,"length"] > 0 && f[2,"length"] > 0 &&

               f[3,"length"] > 0 && f[4, "length"] > 0 && f[5, "length"] > 0 ) {

               allowed[++fn] = substr(line,f[5,"start"],f[5,"length"])

          }

    }

    close(cfile);

}

Here is the 1.0a release... with all the fixes so far as reported...

 

Bug: number of disks in status line counted unassigned slots.   - Fixed.

Bug: slower plug-in speed than in prior releases.                    - Fixed (plug-ins are WAY faster)

Improvement: Added links for lsof and lspci packages to Useful Links in unmenu.conf

Improvement: Added additional parameter passed to plug-ins to allow them to re-create top menu when they supply top of screen heading

Bug: Spaces in file and directory names in file-browser not handled properly. - Fixed.

 

Much thanks to all for the help in testing this.

 

Joe L.

 

Edit: deleted attachment.

It makes a HUGE difference when reading 128 bytes at a time vs 1024..  Pages load MUCH faster... within a second or so instead of 5 or more seconds.

 

I'll say!  Loading most plugin pages went from 8 seconds to near instantaneous on my system.

 

Thanks!

RobJ has submitted a newer version of the syslog plug-in, it needed the same fix to it to deal with syslog files that had a space in their name.  (He helps many people, interpreting messages their syslogs. Apparently, he has saved syslog files with spaces in their names, so he ran into the bug first  )

 

More important to most of us, he has enhanced the color coding of errors in the displayed syslog.  This makes it easier to spot lines in the syslog that may need more attention.

 

RobJ also created a new "Network" plug-in. 

 

I'll include both in future full releases, but until I release 1.0-final, they are in the attachment to this post.

 

Joe L.

 

@ReneV,  lm_sensors requires "Hardware Monitoring Support" support in the kernel... check your kernel .config

 

Also, sensors-detect does require perl.  You can get perl here:  http://packages.slackware.it/package.php?q=current/perl-5.10.0-i486-1

 

However, once you run sensors-detect and know what devices you have and need to modprobe, you no longer need perl.

 

 

It makes a HUGE difference when reading 128 bytes at a time vs 1024..  Pages load MUCH faster... within a second or so instead of 5 or more seconds.

 

I'll say!  Loading most plugin pages went from 8 seconds to near instantaneous on my system.

 

Thanks!

Excellent news. 

 

As I said, thanks for the feedback.  It is really coming together now. 

 

This whole project was to provide a rapid development environment for suggested changes to the unRAID web-interface.  I think it is a success. (It has been rapid, now its even faster )

 

Tom at Lime-technology now has a lot to choose from and think about when he decides to enhance his web-interface.

 

Joe L.

Bug report:

 

If you press the link on the syslog page to download the entire syslog, and then press the "Cancel" button (to not download the file), it brings down the web server.

 

Suggestion:

 

The section that shows the last few lines of the syslog on the main page is not that valuable IMHO.  As an alternative, if it could display the list of currently open files on the array, that would be very useful.

Bug report:

 

If you press the link on the syslog page to download the entire syslog, and then press the "Cancel" button (to not download the file), it brings down the web server.

Is the web-server really down (yes, I see the same error in the browser), or can you just press refresh on the browser?

I can press "refresh" or click on any other link on the top menu and the page will display.

 

Suggestion:

 

The section that shows the last few lines of the syslog on the main page is not that valuable IMHO.  As an alternative, if it could display the list of currently open files on the array, that would be very useful.

let me give some thought how that could be configurable.

 

Edit:  It does not crash when used with Firefox, but does with IE7.  In the interim. don't cancel. 

 

Joe L.

i'm still using IE6.  It crashes (or at least ends) the awk session.  I have to restart it in order to get the web page to redisplay.

 

I was rethinking the idea of having the list of open files display.  It might create some "privacy" type concerns for some users, although I personally would find it useful.  If we could show a list of real errors from the syslog, that would certainly be helpful.  I will try and think of other options that would fit there.