largest usable hard drive in a Series2

[RandC]

I think I've read about this before. What is the process for this? Can you get away with it as easily as deleting two adjacent partitions, and recreating one that covers the same space?

I found this post for starters, had to locate edit commands and procedure. I don't know if it has to be the same space as the deleted partitions, but I did add up what was deleted and made that the new block size for the new partition.

[crashHD]

I think I read that once upon a time, as well. I'm not sure, but I think I also read something about that breaking compatibility with mfstools. I may try that sometime if I need it. Thanks for posting that.

How does one edit the MFS superblock?

[RandC]

How does one edit the MFS superblock?

I use WinHex as my Hex editor

[crashHD]

It worked. I'm up and running.
The whole process, from the beginning:

Installed 3.1.5 instant cake image onto a 2TB drive (this is an image that installs with only a single MFS partition pair).
Installed Series2 SD loopset, 6.2a slices, and upgraded from 3.1.5.
Manually created mfs partition pair with pdisk, with a media partition [strike]2147833648[/strike](presently revisiting this project, and have come to suspect that number may be a typo...do not use until confirmed (6-24-2012)) sectors in size (1 block less than 1TB when using an r value of 2). If you create a precisely 1TB partition, system crashes.
Filled out the rest of the drive with a standard mfsadd -xr2 command.
System info screen cannot handle it, says "0 hours" space. Tivoweb plus info screen also cannot handle it, so now I will have to turn on suggestions just to use them as a free space indicator. System has been running since last night, and is steadily filling the drive via MRV from the other tivo with the sick hard drive. Yay!

That's it. 2TB in a Series2 with a single drive. I used a SATA/PATA adapter from weaknees ($$, but that way was sure to be one troublefree), and a WD EVDS 2TB drive. I don't know if it would have mattered, but I was especially careful to avoid an advanced format drive. I don't know if 4k sector size would have been a problem, but the equation already had enough variables in it...

The hardest part of this process was upgrading from 3.1.5. I must have tried at least a dozen different ways to get that done, before I finally remembered, that loopset slice from back in the day came double gzipped. DOH! It would have been nice if dbload would choke on that, rather than blindly appearing to work, and yet the software mysteriously does not appear in MFS.

2TB is basically the limit. Apple partition map will choke on + 2 TiB (~2.2TB, correct?), so this is (within reasonable rounding) the current limit of a single drive system, right? Any bets on whether or not 4TB is possible with a dual drive setup?

[RandC]

I think I also read something about that breaking compatibility with mfstools.

mfsLive and WinMFS restore with a zone map error so they DO NOT restore properly. The partitions restore the same but something doesn't work out right. I tried mfs_info by changing partition sizes and got a correct CRC but still got a zone map error.

I am getting an error from the info module as well.

-------INTERNAL SERVER ERROR------ --cut here--action:info, path:/details, env:, code:'POSIX EINVAL {invalid argument}'
error during seek on "file91": invalid argument
while executing
"seek $fdDevs($x) [expr 512*$pZone+32]"
(procedure "get_space_tcl" line 35)
invoked from: "get_space_tcl"
(procedure "space_details" line 3)
invoked from: "space_details"
(procedure "info_Space" line 9)
invoked from: "info_$p $chan"
(procedure "MOD::action_info" line 4)
invoked from: "$cmd $chan $path $env"
--cut here--

as you stated now I don't know how much space is left.

[crashHD]

I may have spoke too soon. I've had this tivo copying shows via mrv from an older 500GB unit that was on it's last legs, and it's had a few reboots, and now at a GSOD. I do not know if the issue is because of 2TB, or possibly due to other hardware issues. This system was suffering from random reboots/periodic freezes before the new drive/new OS image.
If someone can't point me toward what log to look to for reboot causes, that will be helpful. Otherwise, the best I can do is wait and see if the GSOD clears, and try swapping the hard drive to another box.

[RandC]

If someone can't point me toward what log to look to for reboot causes, that will be helpful.

I believe the kernel log has some information for you. I would suggest using serial bash, that way you can monitor what is happening while the unit is booting up. I see the error on the laptop monitor so I really don't pay much attention to the log files.

[crashHD]

I don't have serial access.

The unit completed it's GSOD, and now appears to be running alright.

I just now realized Fry's sent me the wrong hard drive. They sent me a EURS, which is an advanced format drive (4k sectors), where I had ordered a EVDS drive. I'm presently on the phone with fry's, attempting to educate this particular service person on the difference between the two drives. I think he thinks I'm a nut.

[crashHD]

I don't think it's related to the 2TB expansion, or the advanced format replacement drive. I think there's something flaking out in the hardware. The periodic momentary freezes are being recorded into the recordings. Playing them back gives momentary glitches and freezes, and playing it back over and over shows the same defects at the exact same time during playback.
I've got an old non-RID box somewhere, I'll be able to swap to that and try it, and process of elimination should tell me if it's the access card or the tivo hardware.

I tried to read the kernel log. Understanding that is beyond my capabilities, although if my theory is right and the problem is in the dss hardware anyway, I don't know if it would have been productive.

[crashHD]

unshelved this project...trying again.

turned out the box used in the first go-round had a power supply that would later fail on me...t'was likely the cause of my intermittent issues...

more to follow...

[crashHD]

Much more success this time. I attribute most of my troubles on the first attempt (a year ago) to a failing power supplyI will recap, from the beginning.

Objectives:
1) Configure a working drive for a Series2 DTivo, fully utilizing a single 2TB hard drive.
2) Fill hard drive via mrv from other less "well endowed" tivos in the house.
3) Preserve ability to put this drive into an HR10, and have full function of it's OTA HD capabilities.

I started with an instantcake image of v3.1.5. I installed it without expanding. Installed this image using an old copy of a beta MFSLive CD that could make sure root partitions were 256MB, kernel partitions the full 4MB, and has an option to expand bootstrap partitions. Not certain what I'll do with that space, but now it's there.

I chose an r value of 4 for the expansion partitions (16MB block size). Something (I don't know what...tivoapp, linux, apple partition map, kernel, whatever) will choke on an MFS partition of >=1TiB (Tebibyte, what a Terabyte would be if hard drive makers measured the same as memory makers...they don't, they all want to make their dicks to look bigger). I'm not sure what it is, but it doesn't matter, the result is the same. A partition on or over 1TiB=Tivo sadness.

So, I manually created a MFS app partition of size 512MB, and manually created a MFS media partition of size 2,147,450,880 blocks. That is precisely 16 Megabytes (one block using an r-value of 4) less than 1 Tebibyte. I don't know if it's necessary for MFS partitions to be sized exactly in increments of it's r-value, but I figure better safe than GSOD.

Then I mfs add that partition pair, and manually made another. I made another 512MB MFS app partition (what goes in these anyway), and a 800 GB MFS media partition. That leaves about 33GB unallocated at the end of the disk. Not sure why I did that, but if I find a use for it down the road, it's there, and if not, it's not like I wasted that much of the drive.

That gives me a partition table that looks like this:

Partition map (with 512 byte blocks) on '/dev/sdb'
 #:                type name                          length   base       ( size )
 1: Apple_partition_map Apple                             63 @ 1         
 2:               Image Bootstrap 1                   262144 @ 64         (128.0M)
 3:               Image Kernel 1                        8192 @ 262208     (  4.0M)
 4:                Ext2 Root 1                        524288 @ 270400     (256.0M)
 5:               Image Bootstrap 2                   262144 @ 794688     (128.0M)
 6:               Image Kernel 2                        8192 @ 1056832    (  4.0M)
 7:                Ext2 Root 2                        524288 @ 1065024    (256.0M)
 8:                Swap Linux swap                   4194304 @ 1589312    (  2.0G)
 9:                Ext2 /var                         2097152 @ 5783616    (  1.0G)
10:                 MFS MFS application region       1048576 @ 7880768    (512.0M)
11:                 MFS MFS media region             2097152 @ 8929344    (  1.0G)
12:                 MFS MFS app region by MFSLive    1048576 @ 11026496   (512.0M)
13:                 MFS MFS media by MFSLive      2147450880 @ 12075072   (1024.0G)
14:                 MFS 3rd MFS app region           1048576 @ 2159525952 (512.0M)
15:                 MFS 3rd MFS region            1677721600 @ 2160574528 (800.0G)
16:          Apple_Free Extra                       68733040 @ 3838296128 ( 32.8G)

Device block size=512, Number of Blocks=3907029168 (1.8T)
DeviceType=0x0, DeviceId=0x0

At this point, I slice upgraded to 6.2a, so I could use MRV to fill this drive from the other tivos it will soon be replacing. That was a week ago. Now there is about 800 hours of content on this unit. System Info screen cannot handle it, and it says it is a "0" hour dvr. Info screen on TWP also cannot handle it. It crashes out with an error before it gets to the bottom of the page where it tells about drive usage. I'm done filling this drive using the mrv functionality, and will be updating this drive to 6.4a at my next opportunity. Then I can install RemoteInfo, and hopefully that one will give me a reliable indication of space used.

This drive is a traditional 512-byte sector size hard drive. Once it gets good and full, the next project is going to be to copy it over to an identically sized advanced format drive, and observe performance and/or issues.

[unitron]

I'm pretty sure I've seen other reports of DirecTiVos reporting 0 hours after a drive upgrade and I don't think all of them were yours.

[crashHD]

The one this 2TB monster is replacing had dual 750GB drives. It showed 0 in the system info screen, but would at least show some garbled info in TWP. IIRC, it showed free space and used space correctly, but for some strange reason came up with a negative number for total space. RemoteInfo does a fantastic job of displaying space stats in the NPL, and it works great with 6.4a (but not 6.2a)

[RandC]

crashHD,
you can make partition16 an type Ext2 and put all of your hacks in there and you don't have to worry about them getting wiped from /var. Just have a check in rc.sysinit.author file check for /var/hack existence, if not then remount partition16 to /var/hack. See below for what I have done. I am using partition7 to currently store my hacks, I don't think I have to worry about anymore SW updates that would swap partitions, and if there were I would do it manually anyway.

#make sure our hacks are always around....
echo "********** $(date) Reboot has taken place" >> /var/log/hack.log
if [ -d /var/hack ]
then
 echo "$(date) /var/hack OK." >> /var/log/hack.log
else
 echo "$(date) /var/hack mount point missing, recreating." >> /var/log/hack.log
 mkdir /var/hack
fi

#mount /dev/hda7 containing hacks to /var, so if /var gets wiped they will alway be there
echo "$(date) Mounting /dev/hda7 at /var/hack" >> /var/log/hack.log
mount /dev/hda7 /var/hack

[RandC]

I'm pretty sure I've seen other reports of DirecTiVos reporting 0 hours after a drive upgrade and I don't think all of them were yours.

I think once you get around 1TB of storage you get the 0 hours reported on the system info screen. Mine has 0 hours as well so there is not an issue with the drive, just the SW.