stuck at few more seconds please

[adrianeaton]

I've got a hacked tivo that will not get past the few more seconds screen. Working fine up until a couple of days ago and then stuck regardless how much I reboot it.

I have tried removing the disk and putting in an old pc on the secondary bus as the master drive and tried a full samsung disk scan which was fine. Also tried renaming the modemtest file in tvbin and that made no difference.

There is one odd thing though. I keep looking at the tverr and tvlog files and they never get any new entries past may31st 00:40. Which I think is when the tivo must have died. Why don't I get any more entries each time it tries to boot?

tverr log shows "thread died due to signal 11" and then "invoking rule 834 rebooting system" any ideas what these mean?

I also tried holding the pause key and trying code 57 and 58 on boot up. 57 didn't seem to do anything and 58 gave a message about software update and after an hour of no screen change I gave up on that.

Can anyone please help?
Adrian.

[Pete77]

Sorry can't help but I suspect a member called blindlemon can probably offer you some more advice on this situation.

Hopefully he will be along shortly.

[blindlemon]

Have you tried wiping the /var parition completely in a PC?

You can use mke2fs -c to clear and recreate it. The TiVo OS should then do the rest at the next reboot. I saw a drive a few weeks ago that had some kind of corruption in the filesystem on /var that was screwing up the /log directory and causing a reboot shortly after "almost there..." - so I guess this could be similar.

Other than that, I would suspect you have some serious MFS corruption that can't be fixed by a GSOD. Have you tried telnetting into the box while it's hanging and having a poke around?

[adrianeaton]

I have tried telnet but it can't connect.

I will try removing the /var as you suggest and see what happens.

Typical it would happen just after I come back off holiday and it is full of stuff i wanted to watch.

I find it a bit strange that there is nothing written to the logs at all? It does get past the first booting screen and so I'd have thought there would be something there with the current date next to it.

Thanks for your help so far.
Adrian.

[adrianeaton]

Is the hdx9 partition the one used for /var?

So to use mke2fs I would type:

mke2fs -c /dev/hdc9

Currently I can see hdc4 and hdc9. hdc7 doesn't want to mount. As I understand it tivo switches between hdc7 and hdc9 as a fallback incase there is a problem? Currently my tivo must be mounting hdc9 I guess.

Cheers,
Adrian.

[blindlemon]

Yes, 9 is the /var - you can check by mounting it :)

[adrianeaton]

deleteing /var didn't seem to make any difference it still gets stuck on the couple of seconds screen.

mounting hdc9 again and I can see it has recreated the standard directories including log. Looking into the log directory and it is empty.

Is there any way of forcing a mfsassert to check the file system? I don't think the panic code 57 seems to be doing anything. I don't get a GSOD just the couple of seconds screen as usual.

How do I see the echo statements in rc.sysinit as they happen or even in a log file in order to debug it?

Cheers,
Adrian.

[Pete77]

Is there any way of forcing a mfsassert to check the file system? I don't think the panic code 57 seems to be doing anything. I don't get a GSOD just the couple of seconds screen as usual.

I believe its mfsassert please from something I remember reading on here.

However perhaps best to wait for our resident expert, blindlemon, to confirm.

[adrianeaton]

I can't get a bash prompt via telnet so it needs to be a command I can run from a pc running a tivo boot cd. Mfsassert doesn't seem to be available on the cd I have.

Adrian.

[adrianeaton]

just tried the old original tivo hard drive before the upgrade and it works fine so definitely the data on the disc is corrupt somehow.

Is there anyway to reset all the files except the recorded programs? or copy off the programs and reset the disk from an image and then copy them back on again?

Thanks for your help
Adrian.

[Pete77]

just tried the old original tivo hard drive before the upgrade and it works fine so definitely the data on the disc is corrupt somehow.

Is there anyway to reset all the files except the recorded programs? or copy off the programs and reset the disk from an image and then copy them back on again?

How old is this drive you are having problems with? I notice that for some strange reason it doesn't seem to be anywhere on your list of possible scenarios that this hard drive has now reached the end of the road and needs to be replaced. Why would you want to keep on using a drive that has clearly already experienced a physical failure. :confused:

The only way you can save programs on a corrupted hard drive is with dd-rescue.

blindlemon is the expert on such matters.

[adrianeaton]

I have done a full scan of the drive using low level samsung HD utils and the physical drive is fine, its not that old anyway. I think it is just the software that has become corrupt

[Pete77]

I have done a full scan of the drive using low level samsung HD utils and the physical drive is fine, its not that old anyway. I think it is just the software that has become corrupt

OK fair point. Although how old is "not that old" and what size is the existing drive?

Also when a new Samsung HD LD 400GB drive is just £60 delivered from www.komplett.co.uk and comes with a 3 year warranty and your current Tivo drive use is very demanding and mission critical mightn't you be better off retiring this drive to some sort of periodic data storage use, in say a USB hard drive enclosure, and putting a new drive in the Tivo?

Tivo drives don't normally become corrupt on their own so either the original setup didn't give it a big enough swap size or copykern wasn't run or it is physically faulty in some way.

[mikerr]

Some versions of the mfsbackup program don't initialise the swap properly (over 127MB)
so tivo won't be able to fix itself on bootup

I'd boot with the LBA48 CD (http://www.tivoheaven.com/download/ptv-mfstools2-large-disk.iso) and run copykern, which initialises the swap as well as copying the kernel.

Then stick it back in tivo and do 58 and 57 again

[TCM2007]

a new Samsung HD LD 400GB drive is just £60 delivered from www.komplett.co.uk and comes with a 3 year warranty.

I do hope you're getting a good commission. ;)

[Pete77]

I do hope you're getting a good commission. ;)

I wish. :p

Its just that my two current Samsung HA250JCs came from Komplett and they delivered in one day, charged the best price and a person can be reached on the phone if you have a query.

So all in all thoroughly pleasant chaps to do business with and charging some of the lowest prices for large hard drives. :up:

[adrianeaton]

My previous drive was swapped under warranty after about 15months and this drive maybe about 9 months old. My drive is a SP1203N 120Gb which is all the space I need really. The warranty is 3 or 5 years so if it is the drive I could get another one free.

Can I use the lba48 cd on a 120Gb drive or is that just for bigger drives.

When I do a 57 what screens should I expect to see? GSOD?

Cheers,
Adrian.

[Pete77]

My previous drive was swapped under warranty after about 15months and this drive maybe about 9 months old. My drive is a SP1203N 120Gb which is all the space I need really. The warranty is 3 or 5 years so if it is the drive I could get another one free.

It would be fairly rare for a drive to suddenly develop problems at only 9 months old.

Perhaps the issue is with how it was originally configured and the swap file size used.

Did you follow the guide at www.steveconrad.co.uk/tivo/upgrade3.html

[blindlemon]

Unfortunately drives can develop bad sectors at any time, although it is less common in a new drive.

Can I use the lba48 cd on a 120Gb drive or is that just for bigger drives.Yes, but you don't need to. However, Mike's suggestion of running copykern is a good one as it will re-initialise the swap header if it has been corrupted and no swap will cause the GSOD to fail.
When I do a 57 what screens should I expect to see? GSOD?You may see the GSOD or "installing TiVo Software" - it varies.

If you can't telnet to the machine when it's in the hung state that implies the hang is happening before the network drivers/telnet daemon are loaded. Can you ping it?

[adrianeaton]

I ran copykern and it completed fine. Tried the HD in the tivo and made no difference still stuck on the few more seconds screen. Tried 57 and 58 paniccodes and no difference.

I can't ping or telnet to it when it is on.

If you don't have any more suggestions I have read how to make a serial cable and analyse the boot up commands to see where it is getting stuck. Is this worth doing? anyone tried it?

Cheers,
Adrian.

[blindlemon]

If you can edit rc.sysinit on your PC then you could try moving the network and telnet startup conmmandssource /etc/rc.d/rc.net
/bin/bash </dev/ttyS3 >& /dev/ttyS3 &
/sbin/tnlited 23 /bin/bash -login &
/sbin/tivoftpdwhich are normally right at the end, to before the main processes are started by the event switcher at# Run the Phase 3 (MyWorld is starting) scripts
runme phase3

cd /var/tmp

# Launch MyWorld and other services (apg, mcp, tcphonehome, dbgc) by telling
# the eventswitcher to launch it (via "switcherstart -l").

if [ "$handcraft" = true ]; then
echo "Running without services."
else
echo "Starting Services."
switcherstart -l
fiIf the problem is in the main processes then you may be able to get network access after the TiVo has hung by doing this.

[adrianeaton]

ok I have tried moving the telnet commands to where you suggested and also to just before the test.conf if statement but neither place gave me a ping response or a telnet.

I tried removing them completely and still at few more seconds screen.

reading sysinit there is a lot of debugging messages and it talks about telnet and some parameters in test.conf, has anyone firgured out how to utilise any of this debugging?

Makes me wish this was windows, I could just recover the video files that were there and then re image the drive and it would be fixed.

any other ideas?
Adrian

[ColinYounger]

Adrian:

Thought I would just answer this:

tverr log shows "thread died due to signal 11" and then "invoking rule 834 rebooting system" any ideas what these mean?

Signal 11 is a program crashing (a signal 9 is a forced death). Rule 834 says that anything crashing which is accessing the MFS database is bad for the TiVo and it should reboot.

So, something crashed and TiVo rebooted to try to recover.

reading sysinit there is a lot of debugging messages and it talks about telnet and some parameters in test.confI expect they would be useful, if we knew where they went.

[adrianeaton]

the debugging messages are just echo'ed. Presumably this goes to the console? The test.conf file looks like it should have parameters for initialising a network connection.

Is rc.sysinit the first file that is run? if not what is there before it? maybe it is getting stuck before the telent service gets started?

Adrian.

[blindlemon]

AFAIK rc.sysinit is the first file run. I don't know anything about test.conf though - sorry :o

[ColinYounger]

I suppose one possibility would be to add ">>/var/BootLog.log" to all the echo's in rc.sysinit, but that assumes /var is mounted. It also assumes that you are confident enough about editing the file correctly, as one teensy mistake trashes the TiVo.

Then you'd be able to see what's happening.

Obviously if you try this, MAKE COPIES OF THE FILES BEFORE EDITING!

[Pete77]

I suppose one possibility would be to add ">>/var/BootLog.log" to all the echo's in rc.sysinit, but that assumes /var is mounted. It also assumes that you are confident enough about editing the file correctly, as one teensy mistake trashes the TiVo.

Surely it doesn't trash the physical Tivo but only the data contents of that hard drive, which at the present time already appears to be trashed? So nothing to lose surely?

[blindlemon]

I think what he means is that it may not even start up to the point it's getting to now.

I suppose one possibility would be to add ">>/var/BootLog.log" to all the echo's in rc.sysinit, but that assumes /var is mounted. You could always put a command to mount /var right at the beginning of rc.sysinit so the log would be written.

Alternatively, write to a log in the root. That has to be mounted to read rc.sysinit ;)

[TCM2007]

Alternatively, write to a log in the root. That has to be mounted to read rc.sysinit ;)

Read only, though.