Bugzilla – Bug 7293
Can't always reconnect after power save mode - conflict with software update?
Last modified: 2009-09-08 09:23:42 UTC
Seems like this might also be interrelated to software update availability as well. I'm not sure how to sort out all of this, exactly, and could use some guidance here. Often, lately, I seem to be getting stuck in the "please wait" connecting to main (my player name). It doesn't finish connecting, ever. I left it connecting overnight last night (in the cradle) and it wasn't connected by the morning. The network strength shows maximum bars in all locations of the room I use the controller in. At other times I was gettting the blue icon and had to manually pick a player to reconnect - this was occurring more frequently until lately when it appears to have switched to this new scenario of "waiting". To make things a bit more confusing: Often powering off/on (in response to loosing connection and not being able to recover it) the controller shows the menu with just a few options for a few seconds before going back into the "please wait" message. If it gets into this situation then it repeats of the power off/on cycle continues to have the problem until the battery is pulled. When it is replaced then it finds software to update and then, after that, manages to connect. The controller version was from Tuesday and the SC version was from Wednesday. I've no idea why the controller isn't updating daily - another question that I've asked upon occasion. I've updated both the controller and SC today to latest available today. Any ideas on how to break this down further? What additional information can I gather? I'd have to think that this would be considered a fairly significant stability problem even though, when released, there won't be many software updates available all that often (if that is the cause, that is).
Thanks for the report, Doug. Our QA team will investigate. I can tell you that the reason that you haven't gotten an update in a few days is that we are in the last days of development and are releasing individual release candidates.
Doug, it would be really useful if you could capture the debug log on your Controller. You can do this using an SD card. First on your PC create a /log folder, then put this card in your Controller. After the Controller has been stuck on "Please Wait" for two/three minutes, turn if off (or remove the battery). You should find a messages file in the /log folder, please attach that file to this bug. Thanks.
Created attachment 2991 [details] Log file I'm not sure this is the same thing Doug is seeing, and indeed I'm not sure exactly how this happened. I came back to my desk and picked up my Jive and it came to life and was spinning, so I put it back down expecting it to reconnect in the usual ~20sec. However, it never did.
Looking at the bug, it seems to think it is connected to a server called DRHODES-X61 or something like that. That server does appear on our network, but I have never connected to it. Mysterious.
SD Card in. I already had a /var/log and /var/core from another debug run. I left those folders in place and added /log. Now waiting for it to show itself again.
I wasn't able to get it to fail last night. I reduced the power save time to 60 seconds and tried many times, but it didn't have a problem. Card is still in and I will try again tonight.
With the log Chris posted Jive never actually went to sleep. Before entering suspend mode Jive will disconnect from all SqueezeCenters. The problem in Chris's case is: 1. Jive has not been connected to two SqueezeCenters (NB-DRHODESX61 and Installs-Duet-MP-FW.local) for over two minutes. (It will connect to all SqueezeCenters to discover players, for example in Choose Music source). 2. After all servers have been disconnected Jive restarts it's server discovery. This deletes any servers that have not been seen for two minutes. 3. The server discovery detects these servers again, and reconnects to them. 4. Steps 2 and 3 occur while Jive is waiting to confirm it has disconnected from all SqueezeCenters. As it has reconnected after the call to disconnect it is stuck. This will be correctly fixed as part of bug 6683. In this case Jive will always be scanning for SqueezeCenters (every minute or so), so will not erroneously delete the state of these SqueezeCenters. Running with two SqueezeCenters here I can't recreate this bug, as it's a race condition between detecting all SqueezeCenters are disconnected and erroneously reconnecting to some SqueezeCenters. This will only happen for people running two or more SqueezeCenters, and I suspect you would be unlucky to see this with only two. A quick fix for 7.0 would be to apply the patch below. This will force disconnection while waiting for all SqueezeCenters to be disconnected. This is probably not the same bug that Doug is seeing, but it will effect people on the MV network and make QA's task of recreating Doug's conditions harder. Index: jive_squeezeboxjive/share/applets/SqueezeboxJive/SqueezeboxJiveApplet.lua =================================================================== --- jive_squeezeboxjive/share/applets/SqueezeboxJive/SqueezeboxJiveApplet.lua (revision 2058) +++ jive_squeezeboxjive/share/applets/SqueezeboxJive/SqueezeboxJiveApplet.lua (working copy) @@ -934,10 +934,10 @@ -- disconnect from all SlimServers if slimServers then - slimServers:disconnect() - local connected repeat + slimServers:disconnect() + connected = false for i,server in slimServers:allServers() do connected = connected or server:isConnected()
I may be closer to this cause than thought. I actually do have 2 SCs running in my network. One of them is current and the one that I'm using the other is a VERY OLD version 7 on linux (SlimServer Version: 7.0a1 - 11737 - RedHat - EN - utf8). I've been waiting for the final build of the Instant SqueezeCenter to reformat/upgrade it (I'm extremely rusty with Unix/Linux and had less time than I thought to dive back into it). I've been keeping it running for backup and disaster failover purposes while running the latest SC version on the primary one. Sorry, I never even thought to mention it before. I didn't realize that it was trying to connect to all SCs on a network, although in hindsight, that does seem pretty obvious.
Patch applies in r2062 for testing.
ross & chris to try to reproduce
I tested this 15 times and a twice there was a brief delay before Jive reconnected (blue icon), about 5 seconds once and as much as about 30 seconds the second time, but every time Jive successfully reconnects. r2065. I will keep an eye out for this.