49 replies to this topic

[dfdream]

All.

I have noticed recently an increase in screen freezes on my 7020 with Jade.

It happens when I change channels.

The screen locks up but I can telnet in and do shutdown now -r to restart.

One in five times this doesnt work and I need a hard reboot.

Is this type of issue related to in adaquate hardware or buggy software.

Sometimes I can change channels or channel hop no problem.

But recently I have lots of freezes.

The only obvious thing I have done is add extra channels to the tv times UK epg download.

I have used lots of satellite receivers and never seen any act like this.

Its the only flaw in a great device.

Any one have any advice to improve the channel hop freeze issue on 7020

dR

[RubbeRNL]

I have a similar problem. Upon pressing the up or down channel to see the channels in the bouquet Im in, this sreen freezes.
The freeze can last for 10 seconds, but also for several minutes or longer.

What seems to help for me is changing the timesync from transponder to system time.
This can be done in the system settings menu.

[WanWizard]

We have a feeling it is related to SQLite. Currently I've been the only one reporting this issue (on a 7000), if you to can confirm you're both using SQLite?

[dfdream]

Yes I use SQLite database for the SKY UK EPG.

Freeze on updating channel list varies from 5 secs to indefinitely. Requiring a reboot.
Some times it allows telnet to do a shutdown now -r and sometimes even telnet is not possible. All times when the freeze occurs web browsing to device doesnt work.

D

We have a feeling it is related to SQLite. Currently I've been the only one reporting this issue (on a 7000), if you to can confirm you're both using SQLite?

[WanWizard]

On the 7000 I haven't had it that bad, it always recovers, although it sometimes takes a minute. And I can always telnet and kill enigma.

The WebIf is run by enigma, so if a thread is blocked somewhere, it's logical that the WebIf doesn't respond either.

[dfdream]

Wan Wizard.

Thanks for your reply.

In your experience what would you say is the cause of this issue. My background is PCs/Windows and would normally see things like this with memory leaks.

When I have a freeze and reboot the DB I might have the same issue again in 5-10 mins or it might go for a few days.

I have noticed (initally anyway) a slight improvement and the freezes sort them selves quicker since installing CCCam 2.10.

But even as you said with yours to sometimes to have to wait 5 secs to a minute for a screen to update is pointing to some problem.

If a provider was having this issue they would not be in business long would they ?

I know that the device is software based but even PCs with Linux or windows are more reliable.

Im surprised more people havent commented on this and wonder do they just accept it or could it be that the hardware I have is problematic.

Also when you say kill enigma is this a reboot using shutdown now -r or is there a command to just restart enigma not requiring a full reboot.

And is there any other way to recover remote control of the device rather than turning off device or telnet.

Thanks


D

On the 7000 I haven't had it that bad, it always recovers, although it sometimes takes a minute. And I can always telnet and kill enigma.

The WebIf is run by enigma, so if a thread is blocked somewhere, it's logical that the WebIf doesn't respond either.

[WanWizard]

I think the CCCam issue is an effect, and not a cause. Either a thread isn't released timely, or there is indeed a memory leak somewhere. And I think the SQLite code/implementation is the cause of all this.

The freezes (not the GUI, but the image) are caused by a CCCam process crashing. You can check if that's the case by checking your entitlements. On a SKY card you receive a lot of ECM's, after a day you should have received hundereds. This counter is reset when CCCam restarts its process.

Unfortunately, the last time I programmed in C was 20+ years ago, and I don't know anything about the SQLite implementation either. But I've requested a debug version of the enigma binary, with lots of logging statements related to SQLite actions, so I might be able to figure out what the problem is. I guess we have to wait until one of the developers has the time to create this special version.

What I mean with 'killing enigma' is telnet to the box and type 'killall enigma'. This will effectively restart enigma, the same as when you would choose 'restart enigma' from the menu. The only difference is that when you kill the process, you lose any config changes you have made, because enigma keeps those in memory, and only saves them with a proper shutdown or restart.

[pieterg]

killall enigma is a 'nice' shutdown, settings are saved. However killall -9 enigma does cause settings not to be saved

[WanWizard]

Ah, ok. Learning all the time...

[RubbeRNL]

Sorry for the delay, but yes, I also use SQlite, and the file is over 20MB. (DM7020).

As said before, in my case, there is a really big improvement when using system time
instead of transponder time.

Just now, I switched back to transponder time, because the system time causes my
box the slowly delay the time ( 3 minutes delay in 2,5 days.). Immediate freezes on
the bouquet/channel list screen. Just zapping back and forward is ok.

After time sync, back to system time, and all is ok.

[WanWizard]

Hmmm, I'll check that out later (when the rest of the house is sleeping ), see if that is the case on a DM7000 as well. Sounds odd though...

[WanWizard]

Interesting.

I've been zapping and flicking through the service selector almost non-stop for the last hour, with satellite time changed to system time.

And I have to say almost no freezing of the GUI, only an occasional hickup.

[dfdream]

All,

I will try this when I go home and let you know.

If set to system time is there a way for the time to be updated periodically to ensure that recordings etc are ok.

Either with some other utility or to have a script to enable transponder time sync during the night for a few mins to update and switch back again to system time.

D

[WanWizard]

The cdk boxes have rdate, for the openembedded boxes you can install the ntpdate package. I've put rdate in cron on my 7000 for the time being.

[dfdream]

Been messing around with it for the last hour with the time set to system time and have noticed a huge improvement.

One had 1 delay for 15-20 secs other than that it seems faster than it ever had before to change channel or bring up the channel list.

Would it suffice to change to transponder for a while to check the time or can rdate be installed on 7020 to run on a schedule.

If yes can you give some details.

I cant believe that the transponder time setting could cuse so much problems.


D

[WanWizard]

Neither can we. We checked our code repository, and there haven't been changed to the EPG code in ages. But the facts don't lie...

For an openembedded box like the 7020 you need ntpdate instead of rdate, see http://www.pli-image...&thread_id=6970 for instructions.

[RubbeRNL]

I installed the ntp packages with telnet as followed ;

ipkg install ntpdate
ipkg install ntp-dbg
ipkg install ntp
ipkg install ntp-bin

I'm not sure all are needed, but I installed last night, and so far the time is ok.

You can easily check the times by installing the plugin sysinfo (0.8.7)
It is in the Pli software downloads, under addons/plugins

3rd option is time/date, it checks box time, ntp time and transponder time.
All 3 times can easily be compared.

Just to point out ; be4 Jade I was using Helenite, I did not experience these freezes.
And I used all the same settings, and the SQlite was about the same size, and the same channels.

Anyway, thanks for confirming that this setting can indeed have a big impact on enigma.
Although it doesnt seem to affect evry1 ?

[WanWizard]

It seems to affect everyone using SQLite (which might not be a lot of people, perhaps due to the performance problems).

I can confirm it wasn't a problem in Garnet and Helenite, I noticed small delays in Iolite (but workable), and a disaster in Jade. We haven't been able to figure out the relation between the time source and the freezes sofar, but we're working on it...

[nfnovice]

It seems to affect everyone using SQLite (which might not be a lot of people, perhaps due to the performance problems).
.

I am using transponder time and SQLite for epg on a 600, 7000 and 7020 - so both OE and CDK image

I experience none of the issues you describe here with Jade Final (or earlier versions)

there might be another contributing factor to the mix which causes these symptoms

[WanWizard]

What's the size of your epg.db, and on what device is it stored?