Modify

Opened 7 years ago

Closed 3 years ago

#7925 closed defect (wontfix)

LuCI hangs when applying changes with uhttpd release 15 (r22630)

Reported by: ledwards@… Owned by: developers
Priority: high Milestone: Chaos Calmer 15.05
Component: packages Version: Trunk
Keywords: uhttpd luci Cc:

Description

Running the latest trunk and Backfire builds on IXP4xx, the LuCI Network -> Interfaces -> WAN and Network -> Wifi -> Wifi0 pages will "hang" when applying changes via the "Save and Apply" button. The changes are accepted, but when the page refreshes, the "Applying changes" message box is displayed indefinitely. The Firefox status bar displays "Transferring data from 192.168.1.1..." indefinitely. Reloading the page doesn't help (still hung), but selecting another page and then returning to the original page works (changes were accepted). This behavior is consistent and repeatable.

To troubleshoot, I re-flashed a previous build which had uhttpd release 14. It worked fine. So I updated uhttpd SVN to the next revision after release 14 (r22630) and rebuilt the package. I installed this package (uhttpd_15_ixp4xx.ipk) via opkg and the problem returned. So I believe this issue was introduced with r22630.

Attachments (0)

Change History (21)

comment:1 follow-up: Changed 7 years ago by anonymous

LUCI SUCKS!!

Use the command line...

comment:2 Changed 7 years ago by anonymous

LuCI is clean and structured, not like the other mess you find out there!

I can reproduce this behavior if I change something in the wireless tab, like changing antenna setting (madwifi) or when band/channels is changed with a mac80211(ath9k) driver.
But for me this is better than before, then I had do ssh in and restart uhttpd.

comment:3 in reply to: ↑ 1 ; follow-up: Changed 7 years ago by ledwards@…

Replying to anonymous:

LUCI SUCKS!!

Use the command line...

If you don't like it, don't use it! I don't think this is a LuCI bug anyway... do you think web servers suck too?

And if you don't have anything meaningful to add, take your comments somewhere else (like the forums)... instead of posting an anonymous rant!

comment:4 in reply to: ↑ 3 ; follow-up: Changed 7 years ago by jow

And if you don't have anything meaningful to add, take your comments somewhere else (like the forums)... instead of posting an anonymous rant!

Don't feed him.

I am aware of the webserver issue. The problem is that it actually termintes the network connection for a short time to to the network restart in background.

The only clean solution will be serving the page first and then run the service restarts independently on the router.

comment:5 in reply to: ↑ 4 Changed 7 years ago by ledwards@…

Replying to jow:

I am aware of the webserver issue. The problem is that it actually termintes the network connection for a short time to to the network restart in background.

The only clean solution will be serving the page first and then run the service restarts independently on the router.

Uhttpd v14 (r22607) works fine with respect to the problem described. Uhttpd v15 (r22630) and subsequent revisions exhibit the problem... so for me, v14 is better.

It looks like a lot of changes were put into r22630, but I don't think any of the changes made (per the changeset description) would have caused this issue. Therefore, I assumed this was a software bug (and not a design issue), as it worked fine before. Was there a design change in v15 that could have caused LuCI pages to "hang" when the network is restarted?

I think this is a major issue and would like to see it in Backfire 10.03.1 (final) if possible.

comment:6 Changed 7 years ago by chillifire @…

I can confirm this issue also exists with brcm-2.4, atheros and ar71xx platforms, not just the reported ixp4xx. It can therefore be assumed to be platform independent. It was tested with uhttpd version 18 on r23871 and still occurred. downgrading uhttpd to version 14 using opkg resolved the issues on all three tested platforms.
Hope this helps honing in on the issue.

comment:7 follow-up: Changed 7 years ago by chillifire @…

Has this been resolved with LUCI 0.10?

comment:8 in reply to: ↑ 7 Changed 7 years ago by anonymous

Replying to chillifire @…:

Has this been resolved with LUCI 0.10?

Yes (at least for me). I don't experience the problem in the Backfire branch since the upgrade to LuCI 0.10.

comment:9 Changed 7 years ago by jow

  • Resolution set to fixed
  • Status changed from new to closed

Backfire will ship with LuCI 0.10 which mitigates this problem.

comment:10 Changed 7 years ago by router_sky

  • Resolution fixed deleted
  • Status changed from closed to reopened

This problem still exist in the with Blackfire and LuCI 0.10.

If user refresh the page for Network Interfaces or WiFi in a short time, the uhttpd will hang up.

comment:11 Changed 7 years ago by hanno.schupp@…

I can confirm this is the case on brcm24, atheros and ar71xx

comment:12 Changed 7 years ago by jow

I cannot reproduce it, sorry.

comment:13 Changed 7 years ago by hanno.schupp@…

What is it then that might make the difference? Could little things like a space in the SSID make a difference? It is definitely an issue for a number of people using an image I compiled, so happens on several platforms.

comment:14 Changed 7 years ago by anonymous

Another thought. My original image was absed on 8.09 and luci 0.9. There were a lot of issues transferring to luci 0.10 with missing js files and the likes. Could this be an issue here?

comment:15 Changed 7 years ago by jow

I doubt that it is missing JS files if the rest works. Such hangs can happen if the network connection is lost while the changes are applying

comment:16 Changed 5 years ago by florian

  • Resolution set to fixed
  • Status changed from reopened to closed

comment:17 Changed 4 years ago by jow

  • Milestone changed from Backfire 10.03.2 to Chaos Calmer (trunk)

Milestone Backfire 10.03.2 deleted

comment:18 Changed 3 years ago by anonymous

backfire 10.03.1 / LuCI 0.10.0 - bug still exists

another problem: wired & wireless clients randomly exhibit two DHCP leases 5 seconds apart for the same MAC address - WHY ?

Also, are you saying this UI bug is fixed in 10.03.2? If so, why does the wiki say the latest OpenWrt usable version is 10.03.1 ?

http://wiki.openwrt.org/toh/buffalo/whr-hp-g54

Can I use 10.03.2 or not? If not, why?

comment:19 Changed 3 years ago by anonymous

  • Resolution fixed deleted
  • Status changed from closed to reopened

comment:20 Changed 3 years ago by anonymous

Forgot to say that I have installed samba2-server / samba2-common 2.0.10-6 and snmpd 5.4.2.1-5, maybe this is relevant here?

comment:21 Changed 3 years ago by jow

  • Resolution set to wontfix
  • Status changed from reopened to closed

10.03.x is EOL, it will not receive further updates.

Add Comment

Modify Ticket

Action
as closed .
The resolution will be deleted. Next status will be 'reopened'.
Author


E-mail address and user name can be saved in the Preferences.

 
Note: See TracTickets for help on using tickets.