Bug 7261 - Jive needs ray firmware updating screen at setup time
: Jive needs ray firmware updating screen at setup time
Status: CLOSED FIXED
Product: Logitech Media Server
Classification: Unclassified
Component: Controller
: 7.0
: Macintosh MacOS X 10.4
: -- normal (vote)
: 7.x
Assigned To: Ben Klaas
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2008-02-21 08:59 UTC by Spies Steven
Modified: 2009-09-08 09:29 UTC (History)
3 users (show)

See Also:
Category: ---


Attachments
Log from jive during setup (16.50 KB, application/zip)
2008-02-21 11:05 UTC, Spies Steven
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Spies Steven 2008-02-21 08:59:31 UTC
Steps to reproduce:
1 Start with a jive fw 1945 and a ray fw 18 in factory reset state
2 Connect your jive and ray to your network, I used standard wireless configuration
3 Choose your music source, notice ray begins update and jive shows updating screen
4 When jive finishes updating jive returns to the choose music source screen
5 If you select your music source again jive will eventually timeout

SqueezeCenter Version: 7.0 - 17630 - Mac OS X 10.4.11 (8S2167) - EN - utf8
Server IP address: 172.19.120.76
Perl Version: 5.8.6 darwin-thread-multi-2level
MySQL Version: 5.0.22-standard
Comment 1 Chris Owens 2008-02-21 09:32:31 UTC
Get a Jive log
Comment 2 Spies Steven 2008-02-21 11:05:15 UTC
Created attachment 2952 [details]
Log from jive during setup

I think I know what might be happening here.  This might be a situation where my test SqueezeCenter is not forgetting my ray soon enough.  So after I downgrade my ray on a different SC and then connect it back to to my test SC it still recognizes the ray and does a different upgrade process that then fails.  Does this sound feasible?
Comment 3 Blackketter Dean 2008-02-21 21:23:54 UTC
Ben: can you take a look at this?
Comment 4 Richard Titmuss 2008-02-22 04:44:11 UTC
Looking at the log I suspect this may be due to dhcp problems. Ben can you try to reproduce the bug to see if it is a real bug, or side effect from a dhcp problem.

Certainly at the end of the log jive has lost it's dhcp connectivity. It looks like a potential race condition where udhcpc has released the lease after just obtaining one:

Feb 21 10:38:39 (none) local0.info udhcpc[1092]: Sending discover...
Feb 21 10:38:39 (none) local0.info udhcpc[997]: Sending select for 172.19.120.217...
Feb 21 10:38:39 (none) local0.info udhcpc[1092]: Sending select for 172.19.120.217...
Feb 21 10:38:39 (none) local0.info udhcpc[997]: Lease of 172.19.120.217 obtained, lease time 3600
Feb 21 10:38:39 (none) local0.info udhcpc[1092]: Lease of 172.19.120.217 obtained, lease time 3600
Feb 21 10:38:41 (none) local0.info udhcpc[997]: Received SIGTERM
Feb 21 10:38:41 (none) local0.info udhcpc[997]: Unicasting a release of 172.19.120.217 to 172.19.113.11
Feb 21 10:38:41 (none) local0.info udhcpc[997]: Sending release...
Feb 21 10:38:42 (none) local0.info udhcpc[997]: Entering released state

Ah, no. Looking at this it appears two udhcpc processes are running. That could be causing some of the dhcp problems in MV. This is where the second dhcp daemon started:

Feb 21 10:38:25 (none) daemon.notice wpa_supplicant[814]: CTRL-EVENT-DISCONNECTED - Disconnect event - remove keys
Feb 21 10:38:25 (none) user.notice root: wpa_action eth0 DISCONNECTED
Feb 21 10:38:26 (none) daemon.notice wpa_supplicant[814]: Trying to associate with 00:0b:0e:4d:84:40 (SSID='SqueezeTest' freq=2462 MHz)
Feb 21 10:38:27 (none) daemon.notice wpa_supplicant[814]: Associated with 00:0b:0e:4d:84:40
Feb 21 10:38:27 (none) daemon.notice wpa_supplicant[814]: WPA: Key negotiation completed with 00:0b:0e:4d:84:40 [PTK=TKIP GTK=TKIP]
Feb 21 10:38:27 (none) daemon.notice wpa_supplicant[814]: CTRL-EVENT-CONNECTED - Connection to 00:0b:0e:4d:84:40 completed (reauth) [id=0 id_str=]
Feb 21 10:38:28 (none) local0.info udhcpc[997]: Sending renew...
Feb 21 10:38:29 (none) user.notice root: wpa_action eth0 CONNECTED
Feb 21 10:38:29 (none) local0.info udhcpc[997]: Sending renew...
Feb 21 10:38:29 (none) local0.info udhcpc[997]: Lease lost, entering init state
Feb 21 10:38:30 (none) local0.info udhcpc[1092]: udhcpc (v1.7.1.svn) started
Feb 21 10:38:31 (none) local0.info udhcpc[997]: Sending discover...
Comment 5 Richard Titmuss 2008-02-22 04:51:20 UTC
On further investigation the dhcp client should be killed after:

Feb 21 10:38:25 (none) user.notice root: wpa_action eth0 DISCONNECTED

Steven can you edit /etc/network/wpa_action on your jive and insert the logger line, like:

if [ "$CMD" = "DISCONNECTED" ]; then 
	# FIXME using ifdown kills the wifi
	/sbin/ifconfig $IFNAME 0.0.0.0
	/bin/rm -rf /var/run/ifstate

	# Make sure udhcpc is killed
        /usr/bin/logger "kill -TERM "`cat /var/run/udhcpc.$IFNAME.pid`
	kill -TERM `cat /var/run/udhcpc.$IFNAME.pid`
fi

Then try to recreate and capture the log. If you could also capture the output of 'ps' from the serial port that would be great. Thanks. 
Comment 6 Ben Klaas 2008-02-22 07:40:53 UTC
I haven't been able to reproduce this, but from Richard's angle of DHCP problems, my guess is that my network environment (particularly my DHCP server) is different enough that I won't see the problem here.
Comment 7 Spies Steven 2008-02-25 09:09:35 UTC
This seems to be fixed with the latest jive fw, I used 2013.  Resolving as works for me and will reopen if I see it again.
Comment 8 Chris Owens 2008-02-25 09:35:05 UTC
Need updating firmware screen for initial setup.
Comment 9 Ben Klaas 2008-04-17 13:05:26 UTC
fixed in Jive 7.0 trunk r2269

QA should test this fix.
Comment 10 James Richardson 2008-05-08 14:23:29 UTC
(In reply to comment #9)
> fixed in Jive 7.0 trunk r2269
> 
> QA should test this fix.
> 

Verified fixed in 7.0.1 - 19325  r2409 - r23
Comment 11 James Richardson 2008-05-15 12:28:17 UTC
This bug has recently been fixed in the latest release of SqueezeCenter 7.0.1

Please try that version, if you still see the error, then reopen this bug.

To download this version, please navigate to: http://www.slimdevices.com/su_downloads.html
Comment 12 Chris Owens 2009-07-31 10:17:34 UTC
Reduce number of active targets for SC