Update MerlinAU.sh

[Martinski4GitHub]

• Added new method to log messages into a file during the F/W Update process. This replaces the previous method that was causing some issues when trying to unmount the USB-attached drive just before the F/W was updated.

• Added code to show the F/W Update notification date on the menu.

• Moved some functions near the top.

[ExtremeFiretop]

So thank you for refactoring the logging solution. @Martinski4GitHub
I woke up this morning to some real progress! (I went to bed last night early, I was too tired to focus).

However as I was concerned, I just tested updating to the latest firmware and post reboot my USB was still: "SDA(1)" as seen below:

Interestingly, if I unmount "SDA(1)" in the WebUI, I am only left with the SDA folder in my mnt location as seen below:

Why does the unmount when flashing not do the same?
I am always left with both SDA and SDA(1) post reboot after the flash.

Either something is still holding on to the logging/the zip file on the USB, or this is some kinda firmware bug we would be working around.

[ExtremeFiretop]

So far there are 2 ways I can re-create this issue for me.

1. Plugin a USB and simply start a firmware upgrade.
   (Upon reboot, the USB has SDA(1) as the mnt point.)

2. In the SSH terminal. Type in: "umount -lf "/tmp/mnt/SDA"
   (Confirm it's unmounted in WebUI, and the mnt point folder still exists in SSH/WinSCP, and then reboot. Upon reboot, the USB is now called SDA(1)

However, if I go to the webUI, and simply click "Safely remove" it successfully unmounts AND deletes the mnt point. Then I can unplug and replug the USB live without issue, it remounts as "SDA" correctly.

[ExtremeFiretop]

And have the same bug doing from 388.5 --> Down to 388.4...

And also 388.5 --> To the newest Alpha 388.6

The only way it works is for me correctly to click "Safety Remove Disk" from the WebUI, and then unplug and just replug the USB.. (Not preforming an upgrade or a reboot) and it comes back up normal.

The only difference I can see so far is ASUS seem to remove the mount point with the unmount from the WebUI, while a regular unmount command alone doesn't remove the mnt point also.

I tried setting my USB mode to USB 2 in the WebUI and it made no difference to the tests.

[ExtremeFiretop]

I tried implementing this code from yesterday again:

    #------------------------------------------------------------#
    # Restart the WebGUI to make sure nobody else is logged in
    # so that the F/W Update can start without interruptions.
    #------------------------------------------------------------#
    /sbin/service restart_httpd && sleep 5
	
	# Check if USBMountPoint is set
    if [ -z "$USBMountPoint" ]; then
        printf "USB mount point is not set. Skipping unmount."
    else
	
        # Unmount the USB drive
        Say "Unmounting the USB drive at $USBMountPoint..."
        umount -lf "$USBMountPoint"
		
		sleep 10

        # Check if the umount command was successful
        if [ $? -eq 0 ]; then
            Say "USB drive successfully unmounted."

            # Check if the mount point directory exists
            if [ -d "$USBMountPoint" ]; then
                Say "Removing the mount point directory..."
                rm -r "$USBMountPoint"

                # Check if the rm command was successful
                if [ $? -eq 0 ]; then
                    Say "Mount point directory successfully removed."
                else
                    Say "Failed to remove the mount point directory."
                fi
            else
                Say "Mount point directory does not exist."
            fi
        else
            Say "Failed to unmount the USB drive."
        fi
    fi

	sleep 5

Sadly the same result, upon reboot I had SDA and SDA(1).
It seems no matter what I do, even if the mount point is deleted, that the reboot will attempt to mount it "twice?" and cause this issue as long as the USB is plugged in at the time of boot. (End result is SDA and SDA(1)

[ExtremeFiretop]

Whoa.

So I reformatted and renamed the USB. It's now NTFS format instead of FAT and it's name is SDE instead of SDA.
Everything was setup perfectly. Only one mount point for SDE, it appeared mounted correctly in the WebUI.

I then reboot and SDA comes back!!!! Along with SDE. So I have both SDA and SDE post reboot when before the reboot everything was only working alone with SDE.

So seems something is stuck "remembering" to create SDA on bootup.

[ExtremeFiretop]

Completely removed our script, and now all of a sudden the problem is gone. Upon reboot I only have SDE now and no SDA.

So it is the logging from our script, and not a firmware bug. As i suspected up here:

Either something is still holding on to the logging/the zip file on the USB, or this is some kinda firmware bug we would be working around.

Something in the restart addconjob hook is creating the directory for the mnt point on reboot.

[ExtremeFiretop]

Problem is at line 858:

if [ ! -d "$FW_LOG_DIR" ]
then
    mkdir -p -m 755 "$FW_LOG_DIR"
else
    # Log rotation - delete logs older than 30 days #
    find "$FW_LOG_DIR" -name '*.log' -mtime +30 -exec rm {} \;
fi

This runs everytime, and tries to create the log directory on every reboot, in this case my log directory was on on the USB so on every reboot, it tries to create "MerlinAU/logs" under /tmp/mnt/SDA

[ExtremeFiretop]

@Martinski4GitHub

Opened a new Pull request to solve these bugs once and for all:

#48

Auto-merge should be enabled for when you approve I believe :)

Thanks again for your help and happy holidays!
Joel S.

[Martinski4GitHub]

Problem is at line 858:

if [ ! -d "$FW_LOG_DIR" ]
then
    mkdir -p -m 755 "$FW_LOG_DIR"
else
    # Log rotation - delete logs older than 30 days #
    find "$FW_LOG_DIR" -name '*.log' -mtime +30 -exec rm {} \;
fi

This runs everytime, and tries to create the log directory on every reboot, in this case my log directory was on on the USB so on every reboot, it tries to create "MerlinAU/logs" under /tmp/mnt/SDA

Yep, that would do it. Excellent troubleshooting!!

[ExtremeFiretop]

Problem is at line 858:

if [ ! -d "$FW_LOG_DIR" ]
then
    mkdir -p -m 755 "$FW_LOG_DIR"
else
    # Log rotation - delete logs older than 30 days #
    find "$FW_LOG_DIR" -name '*.log' -mtime +30 -exec rm {} \;
fi

This runs everytime, and tries to create the log directory on every reboot, in this case my log directory was on on the USB so on every reboot, it tries to create "MerlinAU/logs" under /tmp/mnt/SDA

Yep, that would do it. Excellent troubleshooting!!

Thank you!!! I kept the log rotation in the same location but just changed to the logic to if the directory exists, do the log rotations, and removed the make directory command.

And then moved the original function which does make the directory in the set log directory function, it seemed to work but if you know a better place to move it feel free!

[Martinski4GitHub]

Problem is at line 858:

if [ ! -d "$FW_LOG_DIR" ]
then
    mkdir -p -m 755 "$FW_LOG_DIR"
else
    # Log rotation - delete logs older than 30 days #
    find "$FW_LOG_DIR" -name '*.log' -mtime +30 -exec rm {} \;
fi

This runs everytime, and tries to create the log directory on every reboot, in this case my log directory was on on the USB so on every reboot, it tries to create "MerlinAU/logs" under /tmp/mnt/SDA

Yep, that would do it. Excellent troubleshooting!!

Thank you!!! I kept the log rotation in the same location but just changed to the logic to if the directory exists, do the log rotations, and removed the make directory command.

And then moved the original function which does make the directory in the set log directory function, it seemed to work but if you know a better place to move it feel free!

Yeah, I saw that. Only checking for log rotation is fine where it was so it's done every time the script runs.
I just made a couple of changes for better locations where to double-check that the LOG directory actually exists. See PR #49.