broken webif

oneillb

New Member
I have 2x Hdrfox-t2 machines, both have custom firmware set up pretty much identically...

Recently the web interface on one has become broken. below is an image of the home page which doesn't show any icons, just a partial rendering of the notification box.

Trying to hit the lower level pages directly results in the same.

Screenshot_20210804-044057.jpg
I've tried to reinstall webif from the telnet maintenance menu, and also reset the custom firmware environment. Not gone as far as the RMA option yet...but nothing yet has had any effect..

Any ideas what may be causing this? Has the notification log become too long?

Thanks,

Barry
 
What's in the /tmp/webif-error.log file?
You may need to drop to a telnet command prompt to see this as the WebIf is broken - cat /tmp/webif-error.log
If the notification log is the problem, then rm /mod/tmp/notify.log and see if it's better afterwards.

Reinstalling non-broken stuff and clearing caches is pointless.
 
When I reset the custom firmware env, I did reinstall from the webif home page, which gave.me the impression that I'd fixed the problem... When webif was reinstalled the problem with the notification window came back... So it must be a something that survives the reset, or possibly a disk error.

I'll have a look at the webif and notify logs when I get home.
 
Reinstalling non-broken stuff and clearing caches is pointless.
We're talking about a defect which survives having the HDD CF stuff wiped out, so it doesn't strike me as pointless at all. The browser cache survives; anything under /mod doesn't.

possibly a disk error
Certainly you should be running fixdisk as a precaution and to establish a baseline, but I am dubious disk errors would create the history you have reported.
 
Not gone as far as the RMA option yet
The only time it should be necessary to use the RMA option is when it is required to revert to standard firmware with no evidence of the custom firmware having been installed. If the requirement is to clean out all the installed packages and associated data to fix some unknown problem then a "CFW Reset" should be sufficient.
 
Hi all,

The /tmp/webif.log only contains a lighthttpd startup message.

The /mod/tmp/notify.log is only 6k in size...it's not a text file so I renamed it out of the way, and normal service has been resumed.

Thanks for your help...
 
Seems like this file must be sitting on a bad disk sector. The WebIf server-side script webif/include/notify.jim was unable to construct the page from the point where You have pending system notifications was added to the page. The next thing it does is to read the entire file and display it under that heading, and that's where it appears to have crunched.

There should have been something in the error log but that's transient.
 
The /tmp/webif.log only contains a lighthttpd startup message.

That is normal which is why the request was for
What's in the /tmp/webif-error.log file?
You may need to drop to a telnet command prompt to see this as the WebIf is broken - cat /tmp/webif-error.log
The /mod/tmp/notify.log is only 6k in size...it's not a text file so I renamed it out of the way, and normal service has been resumed.
At least you have resolved the problem, if you post the old notify.log we may be able to see why it caused a problem and what could be done to prevent future occurences
 
Because reset only reinstalls software, it doesn't remove data files including logs, databases and other configuration data.
So this is wrong then?:
It does the same thing as RMA but without the final step of installing standard firmware or resetting the RMA flag via the telnet menu.
I took that to mean "do everything RMA would do except set the RMA flag". If it does not in fact do everything RMA does, CFW Reset can't be relied upon to clear faults... and I return to my position of suggesting RMA over CFW Reset as a last resort fix.
 
Hi,

Find attached the notify.log ... it has some text in there, but also some binary data too... which is probably what caused the issues....

The webif-error.log file was empty.

Barry
 

Attachments

  • notify.log
    6.4 KB · Views: 7
The /tmp/webif.log only contains a lighthttpd startup message.
It will if you haven't tried to access the WebIf since reboot, since it's on a filesystem in memory.
You need to restore the notify log, then try accessing the WebIf, then post the error log, then rename the notify log away again.
The /mod/tmp/notify.log is only 6k in size...it's not a text file
It is a text file, or should be. That's probably the problem. Do run fixdisk, as has been suggested.
Because reset only reinstalls software, it doesn't remove data files including logs, databases and other configuration data.
Which is why I wrote the last sentence in post #3.
 
Do run fixdisk, as has been suggested.
It does look like a cross-linked file, so that's a definite.
It doesn't crash my WebIf 1.4.9-6, so I guess you're on a different version (it helps to mention which) or it's the disk fault.
 
From /sbin/modinit:
Code:
reset_all()
{
      banner reset all
      keyf=_RESET_CUSTOM_FIRMWARE_ENVIRONMENT
      raflag=0
      [ -f /var/lib/humaxtv_backup/.rma ] && raflag=1
      for kdir in $bootmod \
          /media/My\ Video \
          /media/drive1/Video /media/drive1; do
              [ -f "$kdir/$keyf" ] || continue
              echo "Found reset trigger file at $kdir/$keyf"
              raflag=1
              rm -f "$kdir/$keyf"
      done

      if [ $raflag -eq 1 ]; then
              echo "Resetting custom firmware environment."
              rm -rf $mdir/*
              rm -rf /var/lib/humaxtv/mod
              rm -rf /var/lib/humaxtv_backup/mod
              rm -rf /var/lib/humaxtv_backup/partition
              [ -d /media/drive1/virtual_disk ] && \
                  rm -rf /media/drive1/virtual_disk
              [ -d /mnt/hd2/virtual_disk ] && rm -rf /mnt/hd2/virtual_disk
              [ -f /mnt/hd1/epg.db ] && rm -f /mnt/hd1/epg.db
              [ -f /media/drive1/epg.db ] && rm -f /media/drive1/epg.db
              [ -d /mnt/hd2/mod ] && rm -rf /mnt/hd2/mod
      fi
}

rma()
{
      banner rma
      if [ -f /var/lib/humaxtv_backup/.rma ]; then
              /sbin/display "== RMA =="
              echo "RMA Mode."
              bootstrap rma.html
              exit 0
      fi
}
reset_all and rma are both called on boot.
This suggests that if reset has failed to delete all the files under /mod due to disk corruption then rma would have the same result since it is the reset_all function that performs the deletion..
 
Last edited:
I'm pleased you cleared that up. It seems to me the "universal cure" should be:
  1. fixdisk
  2. Reset CFW
 
Back
Top