TiVo Community
TiVo Community
TiVo Community
Go Back   TiVo Community > Main TiVo Forums > TiVo Series3 HDTV DVRs
TiVo Community
Reply
Forum Jump
 
Thread Tools
Old 10-19-2014, 11:18 AM   #10021
HerronScott
Registered User
 
Join Date: Jan 2002
Location: Staunton, VA
Posts: 982
Quote:
Originally Posted by dlfl View Post
Thanks but it didn't take long looking at those threads to realize this is something I'm not going to attempt -- too much to unravel there.

I would be interested in just learning whether the 2TB expansion that I already have is 4K aligned, however -- if that can be done using the mfsinfo dump that I have. Here is what I have surmised from looking at the partition map listing:
1. The "length" and "base" are given in units of 512 bytes. Thus 4K alignment requires the base be a multiple of 8 (8 x 512 = 4096).
2. For the purposes of minimizing thrashing due to splitting across sectors, the partitions that count the most are those labeled "MFS Media Region", of which there are three. This is based on the idea that these partitions contain the video data, which comprises the bulk of the workload for the disk.

Based on this premise, not all my partitions are 4K aligned but the huge 1.7T MFS Media Region is. So it looks like thrashing should be substantially reduced.

Does this make sense?

The mfsinfo dump was attached to earlier post #9981 and here is the link to the attachment: http://www.tivocommunity.com/tivo-vb...4&d=1412796823
I'd have to let someone with more experience answer your question. Unfortunately my partition map is going to be slightly different for our S3's since they started with a larger hard drive and for our HD since we expanded it with JMFS (only 14 partitions instead of 15).

Here's the S3:

Code:
Partition Maps
 #:                  type name                            length base      ( size  )
  1   Apple_partition_map Apple                               63@1         (  31.5K)
  2                 Image Bootstrap 1                          8@485364800 (   4.0K)
  3                 Image Kernel 1                          8192@485364808 (   4.0M)
  4                  Ext2 Root 1                          524288@485373000 ( 256.0M)
  5                 Image Bootstrap 2                          8@485897288 (   4.0K)
  6                 Image Kernel 2                          8192@485897296 (   4.0M)
  7                  Ext2 Root 2                          524288@485905488 ( 256.0M)
  8                  Swap Linux swap                      262144@486429776 ( 128.0M)
  9                  Ext2 /var                            524288@486691920 ( 256.0M)
 10                   MFS MFS application region          589824@487216208 ( 288.0M)
 11                   MFS MFS media region             216747008@268617792 ( 103.4G)
 12                   MFS MFS application region 2        589824@487806032 ( 288.0M)
 13                   MFS MFS media region 2           268617728@64        ( 128.1G)
 14                   MFS MFS App by Winmfs                 2048@488395856 (   1.0M)
 15                   MFS MFS Expanded by Winmfs      3418624000@488397904 (   1.6T)
and the HD:

Code:
Partition Maps
 #:                  type name                            length base      ( size  )
  1   Apple_partition_map Apple                               63@1         (  31.5K)
  2                 Image Bootstrap 1                          8@171919424 (   4.0K)
  3                 Image Kernel 1                          8192@171919432 (   4.0M)
  4                  Ext2 Root 1                          524288@171927624 ( 256.0M)
  5                 Image Bootstrap 2                          8@172451912 (   4.0K)
  6                 Image Kernel 2                          8192@172451920 (   4.0M)
  7                  Ext2 Root 2                          524288@172460112 ( 256.0M)
  8                  Swap Linux swap                      262144@172984400 ( 128.0M)
  9                  Ext2 /var                            524288@173246544 ( 256.0M)
 10                   MFS MFS application region          589824@173770832 ( 288.0M)
 11                   MFS MFS media region             137629696@174950480 (  65.6G)
 12                   MFS MFS application region 2        589824@174360656 ( 288.0M)
 13                   MFS MFS media region 2           171919360@64        (  82.0G)
 14                   MFS MFS media region 3          3594446848@312580176 (   1.7T)
Scott
__________________
  1. S3 HD - 2TB - Lifetime (son's)
  2. S3 OLED - 2TB - Lifetime
  3. S3 OLED - 2TB - Lifetime
  4. S1 Phillips - 120GB - Lifetime transferred to S3
  5. S1 Sony - 30GB - Lifetime transferred to S3
HerronScott is offline   Reply With Quote
Old 10-19-2014, 01:32 PM   #10022
jmbach
Registered User
 
jmbach's Avatar
 
Join Date: Jan 2009
Posts: 833
@dlfl No it is not 4k aligned.
__________________
"Delay is preferable to error" - Thomas Jefferson
"If I have seen further it is by standing on the shoulders of Giants" - Sir Isaac Newton
jmbach is online now   Reply With Quote
Old 10-19-2014, 03:27 PM   #10023
nooneuknow
TiVo User Since 2007
 
Join Date: Feb 2011
Location: Cox Cable Market, NV
Posts: 3,043
TCD652160 TiVo HD 2TB 4K aligned (and physical layout optimized)

Partition Table:
# StartSector SizeInSectors Name Type
1 0x00000001 0x0000003F Apple Apple_partition_map
2 0x12735840 0x00000008 Bootstrap 1 Image
3 0x12735848 0x00002000 Kernel 1 Image
4 0x12737848 0x00080000 Root 1 Ext2
5 0x127B7848 0x00000008 Bootstrap 2 Image
6 0x127B7850 0x00002000 Kernel 2 Image
7 0x127B9850 0x00080000 Root 2 Ext2
8 0x12839850 0x00096000 Linux swap Swap
9 0x128CF850 0x00080000 /var Ext2
10 0x1294F850 0x00090000 MFS application region MFS
11 0x0A3F4840 0x08341000 MFS media region MFS
12 0x129DF850 0x00090000 Second MFS application region MFS
13 0x00000040 0x0A3F4800 Second MFS media region MFS
14 0x12A6F850 0x00001000 New MFS Application MFS
15 0x12A70850 0x7FFF8000 New MFS Media MFS
16 0x92A68850 0x563A0060 New MFS Media MFS

Physical Disk Layout:
# StartSector SizeInSectors Name Type
1 0x00000001 0x0000003F Apple Apple_partition_map
13 0x00000040 0x0A3F4800 Second MFS media region MFS
11 0x0A3F4840 0x08341000 MFS media region MFS
2 0x12735840 0x00000008 Bootstrap 1 Image
3 0x12735848 0x00002000 Kernel 1 Image
4 0x12737848 0x00080000 Root 1 Ext2
5 0x127B7848 0x00000008 Bootstrap 2 Image
6 0x127B7850 0x00002000 Kernel 2 Image
7 0x127B9850 0x00080000 Root 2 Ext2
8 0x12839850 0x00096000 Linux swap Swap
9 0x128CF850 0x00080000 /var Ext2
10 0x1294F850 0x00090000 MFS application region MFS
12 0x129DF850 0x00090000 Second MFS application region MFS
14 0x12A6F850 0x00001000 New MFS Application MFS
15 0x12A70850 0x7FFF8000 New MFS Media MFS
16 0x92A68850 0x563A0060 New MFS Media MFS

I suspect the questionable "extra pair" of partitions (physically placed at the end) is due to TiVo never shipping any (TiVo HD) units with a software version that broke the ~1.26TB barrier, and this image having a shipping version of (TiVo HD) software.
__________________
Cisco tuning adapters should never be used inline (using the TA coax OUT port) to connect a TiVo, if MoCA is in use. Use a splitter w/PoE filter on leg to TA, use other leg for the TiVo. Enjoy!

Last edited by nooneuknow : 10-20-2014 at 08:58 AM. Reason: added (TiVo HD) context for extra partitions
nooneuknow is offline   Reply With Quote
Old 10-19-2014, 04:07 PM   #10024
dlfl
Cranky old novice
 
dlfl's Avatar
 
Join Date: Jul 2006
Location: Near Dayton, Ohio
Posts: 6,067
Quote:
Originally Posted by jmbach View Post
@dlfl No it is not 4k aligned.
I understand that not all partitions are 4k aligned but do you agree my large 1.7T partition is 4k aligned based on the base address being a multiple of 8?

And doesn't that mean the bulk of my disk I/O ops will be to that aligned partition, which should minimize splits across sector boundaries?
__________________

"It must be swell to be so perfect and odor-free" -- Del Griffith


VideoReDo users: Try
To view links or images in signatures your post count must be 10 or greater. You currently have 0 posts.

pyTivo users: Try
To view links or images in signatures your post count must be 10 or greater. You currently have 0 posts.
and
To view links or images in signatures your post count must be 10 or greater. You currently have 0 posts.
dlfl is offline   Reply With Quote
Old 10-19-2014, 06:24 PM   #10025
jmbach
Registered User
 
jmbach's Avatar
 
Join Date: Jan 2009
Posts: 833
Yes the 1.7TB partition would be considered 4k aligned. Only if the disk I/O all occurs in that partition. That is the rub. How often the TiVo needs to access the root, kernel, swap, /var, and partitions 10 and 12 during operations is anybody's guess.
__________________
"Delay is preferable to error" - Thomas Jefferson
"If I have seen further it is by standing on the shoulders of Giants" - Sir Isaac Newton
jmbach is online now   Reply With Quote
Old 10-20-2014, 08:02 AM   #10026
nooneuknow
TiVo User Since 2007
 
Join Date: Feb 2011
Location: Cox Cable Market, NV
Posts: 3,043
The way that most commercial (WeaKnees) and ebay pre-imaged drive sellers do things is "aligned and optimized".

"Aligned" handles aligning the partition start boundaries.
"Optimized" makes a sandwich of the physical locations of the partitions, placing them in a manner that lowers the internal workload, with seek operations being the most critical factor, even more so with AF drives.

It takes doing both, in order to maximize drive performance, and lifespan. While the performance improvements may not always be noticeable (in TiVo use), due to the AF/4K/512e Read-Modify-Write operations and seeking related internal thrashing being masked by 64MB of cache (TiVo's stock non-AF drives had the smallest cache WD offered to OEMs), it is well known that less seeking, shorter seek distances, balanced seeking, and less RMW operations, will provide a longer lifespan, than if the same drive was used without alignment and optimization.

Of course, if a drive is just defective to begin with, or develops defects that would ultimately shorten the life of the drive, no amount of anything can stop such defects from ending the drive's life. This inevitably will lead to some naysayers/detractors, who will say that none of any of this matters, because their aligned and optimized drive lasted 6 months, while their as-is drive lasted 6 years...

Given the state of platter drive QC/QA, DOA rates, "infant mortality" rates, and other issues, plaguing the platter drive industry, over what is coming up to nearly a year of bad reviews for seemingly all platter drives (WD seeming to have the blackest eye), it is going to be a long time before anecdotal reports of lifespans can begin to mean anything again.

A factor which makes the "optimization" part hard for many to achieve, is those who want to keep what is on the drive they have, and just copy it to a larger drive, while also adding capacity. The SOP DIY method for this, is to just tack on a huge partition at the end of the existing image. Even if it winds up coincidentally "aligned" at the start boundary, it doesn't change that the drive is now "lopsided" (nor does it align pre-existing non-aligned partitions). That seems to be what some are trying to ask/say here: "My ~1.7TB expansion partition appears to be aligned, so I'm good, right?". We all know it works, which has been "good enough" for most (but, still not a full alignment, nor an optimal physical layout).

Lately, it seems more are looking to align all the partitions, to achieve an aligned drive. Optimizing/balancing the physical partition locations still seems to not be of much interest. It's hard enough just to make the whole drive aligned, for most.
__________________
Cisco tuning adapters should never be used inline (using the TA coax OUT port) to connect a TiVo, if MoCA is in use. Use a splitter w/PoE filter on leg to TA, use other leg for the TiVo. Enjoy!

Last edited by nooneuknow : 10-20-2014 at 08:51 AM. Reason: added context
nooneuknow is offline   Reply With Quote
Old 10-20-2014, 01:40 PM   #10027
_dylan
Registered User
 
Join Date: Oct 2014
Location: Vegas
Posts: 2
So after a couple days fighting the Tivo, I got it sorted out.

I tried to recover the original drive since it looked like there was only one error on the drive it, but after a couple tries it didn't work.

Thanks ggieseke & nooneuknow for the help.

One thing to note is that you'll need to let the tivo update to the current software before doing a Clean & Delete Everything. Then it will do the process again.

So now I guess I'll have to wait until the drive fails again.
But next time the issue will be supper easy to fix.
_dylan is offline   Reply With Quote
Old 10-20-2014, 04:51 PM   #10028
nooneuknow
TiVo User Since 2007
 
Join Date: Feb 2011
Location: Cox Cable Market, NV
Posts: 3,043
Quote:
Originally Posted by jmbach View Post
Yes the 1.7TB partition would be considered 4k aligned. Only if the disk I/O all occurs in that partition. That is the rub. How often the TiVo needs to access the root, kernel, swap, /var, and partitions 10 and 12 during operations is anybody's guess.
Based partially on the use of my ears (and sometimes a mechanic's stethoscope), the disk I/O in some of those partitions is intense, when in the process of indexing and GC. Let's just say that when using a drive with no AAM, and no "intelliseek", you can record all tuners concurrently, and not be able to hear much without a stethoscope, while nothing but your ears are required to tell when indexing and GC occur (and a stethoscope might damage your hearing during those processes).

I would almost wonder if the recent reports that the whole GC not completing issue, is verifiably only affecting larger than stock drives, could be helped by increasing the efficiency of the disk I/O, by reducing internal drive latency, thus reducing resources required to complete GC. Indexing and GC operations, although considered "background processes", are the most I/O intensive, seek intensive, and most susceptible to the issues AF drives brought into the equation, namely "Read-Modify-Write" operations, instead of just "Write" operations, as it would be on a native 512 byte sector drive.

An experiment I propose somebody try (with very good note keeping skills):

1. Find yourself having a TiVo, with an AF drive, that won't complete GC.
2. Clone the drive to a comparable native 512 byte sector drive of same size.
3. Run the clone, and see if GC completes, or completes more often.
4. Make sure nothing else was changed, that may have skewed result validity.
5. Clone in the reverse direction, and repeat #4.
6. Run the original drive, and see if GC completion is negatively affected.
7. Share all the details, notes, and results, with the rest of the class.

It's not a perfect test, and not everybody has a 512 byte sector native drive handy. But, if enough people try this, and the results are conclusive enough for plausibility of non-media partition disk I/O taking a hit on AF drives, enough to degrade Indexing and GC completion, then the importance of physical layout and alignment are made clear. Even if all that can be seen is a significant improvement in Indexing and GC completion times, that's noteworthy.

Since all drives have highest performance on the outer tracks, and follow a downward curve to the lowest performance on the inner tracks, the location of the databases, and other routinely processed non-AV data, with the highest random I/O demands (especially seek-intensive), would ultimately be placed strategically (not necessarily on the outermost tracks). The drives that are used in TiVos, and the ones most use to upgrade them, have good sequential performance. But, it literally falls off a cliff, from whatever sequential performance zone it is in, when it's non-sequential, and the drop starts from whatever point on the curve the drive would be at, if doing sequential operations.

Realistically, all drives should be able to handle the AV data that gets written to the Media partitions, no matter where it is on the platters, or if the media partitions are aligned, due to the large bulk writes. But, I'm not saying alignment has no value there (think fragmentation).

In summation, there's a reason why certain non-media partitions tend to be smack in the middle of the media partition pairs, and why they tend to come in pairs. There was a time when this was critical, then drive technology made it less so (via density and performance), then drive technology (via AF) seems to have made it just as important again, and it would be already clear, just how much, if large onboard drive cache wasn't camouflaging internal AF pitfalls (pulling the wool over our eyes).

If you have already upgraded, and don't have any issues, I don't advise starting over, unless the thought of making the drive work harder than it needs to, keeps you awake at night. If you are planning an upgrade, and want the drive to have the least workload overhead, then this stuff is for you.

Some afterthoughts:

Another potential performance-sapping drive technology that has somewhat invaded most modern drives (suspiciously around the advent of AF), is a replacement of AAM (where you could improve performance and have a louder drive, or have a quiet drive and lower seek performance, or set it somewhere in-between). WD's AAM replacement brand is "Intelliseek". These non-adjustable, and non-disableable seeking algorithms serve two purposes:

1. Quieter seeking (more of a secondary result).
2. Lower wear of the head actuator, and prevention of seek-errors, by preemptively timing seek operations, to make them less snap-action (and make them smoother), and attempt to lower the number of overall seeking operations, by consolidation and queuing of them, possibly even re-ordering them.

What they don't tell you, is that this technology increases latency internally, and is yet another feature that hides its pitfalls by using on-drive cache. I have no proof of this technology being a detriment, outside the drive. But, in theory, an unaligned drive, not optimally physically partitioned, just might overwhelm the processing capabilities for queuing up optimized seeks (drive optimized). I won't write a page worth of explaining how many different ways that could "go sideways".
__________________
Cisco tuning adapters should never be used inline (using the TA coax OUT port) to connect a TiVo, if MoCA is in use. Use a splitter w/PoE filter on leg to TA, use other leg for the TiVo. Enjoy!

Last edited by nooneuknow : 10-20-2014 at 07:37 PM. Reason: typo
nooneuknow is offline   Reply With Quote
Reply
Forum Jump




Thread Tools


Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Advertisements

TiVo Community
Powered by vBulletin® Version 3.6.8
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
vBulletin Skins by: Relivo Media

(C) 2013 Magenium Solutions - All Rights Reserved. No information may be posted elsewhere without written permission.
TiVoŽ is a registered trademark of TiVo Inc. This site is not owned or operated by TiVo Inc.
All times are GMT -5. The time now is 11:21 AM.
OUR NETWORK: MyOpenRouter | TechLore | SansaCommunity | RoboCommunity | MediaSmart Home | Explore3DTV | Dijit Community | DVR Playground |