DSM-G600, DNS-3xx and NSA-220 Hack Forum

mig wrote:

SilentException wrote:

can someone share a lsof binary for dns-323?

http://dns323.kood.org/forum/t2139-lsof-DNS-323.html

thank you
i didnt feel like setting optware so i just got http://ipkg.nslu2-linux.org/feeds/optwa … 1_arm.ipk, renamed to .tgz and took the binary from there
works fine, maybe now i can see why is mldonkey keeping my hdd's up.....

swap file?

used swap space is possible. if you create a swap space on your usb flash drive dont forget to deactivate the swap on the hd's or set the swap space priority on the flash drive much higher. 2nd possibility would be to adjust the kernel swapiness.
A partition is not even necessary. you can also format a file on a filesystem to swapspace and use it. e.g.:
dd if=/dev/null of=/mnt/usbstick/swap.file bs=1M count=512
mkswap /mnt/usbstick/swap.file
swapon /mnt/usbstick/swap.file

Last edited by oxygen (2008-05-13 04:10:44)

I have a couple of questions - so please pardon my ignorance

First - if I understand correctly flash memory has a limited life expectancy, it literally wears out, so much so that solid state disks have special "wear leveling" algorithms that move data around to prevent premature disk failure - wouldn't the use of a flash drive for a swap file have similar "wear" issues.

Second - read/write speeds - maybe this is of little relevance, if the intention is to store files being downloaded from the internet, since the bottleneck is likely to be the internet connection, but as I understand it, write speeds to flash memory are pitiful when compared to a disk.

I guess my real curiosity is why is there this obsession to use flash memory with the DNS-323 when there is ample disk space available?

For myself, it's more about (in no particular order):

1) Be Green.  My DNS-323 runs at 36 Watts at boot, 27 Watts when both drives are spinning, about 18 watts when only 1 drive is spinning, and 9 watts when running off the USB flash drive.

2) Less mileage on the hard drive.  If a HD fails, that's 1 TB lost.  Thumb drive?  At 2 GB for $15 for less, it's easy to pop another one in there if it fails.  Downloading bittorrent or running PHP/MySQL is not the main reason I bought my 2 TB of storage, so I'd rather minimize the regular wear and tear on the HD.  Besides, the 1 GB thumb drive I stuck in the DNS-323 is an old stick I didn't use much any more, so I was happy to commission it for something useful.

3) Silence.  No hard drive spinning, no noise.  And without the fan, the unit is completely silent.  Before this, the hard drive would spin up every 10 minutes or something for logging, then spin down.  I've always thought my Desktop PC was noisy, so this is a nice change.

4) It's fun.  Actually solving all the issues it would take to run everything I want to off the USB flash drive is a technical challenge in itself, although it's mostly solved here in this forum.  It's rewarding to walk by the DNS 323 and hear nothing, knowing I've made everything run on the thumb drive.

Last edited by raid123 (2008-05-13 11:43:40)

SilentException wrote:

...what did you do to lower memory consumption of mlnet. atm mine uses 35-40mb which is definatelly too much (all plugins but bittorent are disabled!)

Were you looking at the output from 'top' or from 'free'?  If it was from 'free' as I supect, then check 'top' for memory usage.  Linux uses available memory like swap space first, because it is much faster than swap space.  It frees it up as programs need it, so at first glance, it looks like your memory is always being sucked up.  It's a more efficient way of doing it than some other operating systems.

bq041 wrote:

True, but a swap file is pretty slow in comparison to a partition.

Thats a common missunderstanding. The difference is not noticeable.
see: http://lkml.org/lkml/2005/7/7/326

if you afraid of a performance hit, create the swapfile before other files on the partition. this will prevent fragmentation (which isn't an issue on flash devices at all).

on flash drives, fragmentation has no impact on performance so i created a 128mb swap file and added it to swap total with swapup. my question is how do you tell kernel the priority so it uses that swap before the disk swap?

SilentException - did you read the link?

blahsome - whilst I generally have some amount of respect for Serdar Yegulalp's contributions, I do not always agree with him - you may notice that he refers to "data that I have seen" rather than tests that I have done - the link that I have provided does show the tests that that individual claims to have done - so for now I'll respond like Serdar Yegulalp and simply say that the data I have seen does show that fragmentation can have a significant impact on flash memory performance.

We have moved from whether or not fragmentation is an issue on flash drives - my position is simply that it does occur - to whether or not defragmentation of flash makes sense and it appears that Serdar Yegulalp's position is "perhaps" ....

In short, defragmenting flash memory is probably not worth it unless you can demonstrate that there is a perceptible speed improvement by doing so. The key word is perceptible, and unless you are using measurable and testable metrics for judging such a thing, you may not be witnessing anything other than subjective bias about how fast such things should be.

Unless I'm very much mistaken what is provided in the page I linked to can be defined as "measurable and testable metrics".

Now - whether or not performance degradation does occur is of little importance to me - it has no impact on my usage of flash memory which is essentially as "portable storage" - if you like a compact, high capacity, diskette - so I'll step back at this point.

SilentException wrote:

on flash drives, fragmentation has no impact on performance so i created a 128mb swap file and added it to swap total with swapup. my question is how do you tell kernel the priority so it uses that swap before the disk swap?

normally this is done via swapon -p pri device. unfortunatly in the busybox version of swapon, this parameter does not exist. i just deactivated the swap on the disks. my 128 mb swap are more than enough and i have quite a few apps running (own mt-daapd, mediatomb, rsyncd, nfsd)

fyi: you get some infos about the swap via cat /etc/swaps.

fordem wrote:

SilentException wrote:

i did read it now. didnt even see there is a link the first time
so you're right, and those benchmarks prove it.. still, if i used dd to create swap on a blank usb, it should be fine and contiguous right?

It'll be contiguous only directly after it's created - swap files are constantly having blocks changed - ie written, read, over written etc., personally I feel that if flash memory is used for a swap file that fragmentation will be unavoidable.

Thats just wrong. Seems like you didn't read the link i posted:

> 2. Is it possible for the swapfile to become fragmented over time, or
> does it just keep using the same blocks over and over? i.e. if it's
> all contiguous when you first create the swapfile, will it stay that
> way for the life of the file?

The latter.  Create the swapfile when the filesystem is young and empty,
it'll be nice and contiguous.  Once created the kernel will never add or
remove blocks.  The kernel won't let you use a sparse file for a swapfile.

Oxygen - no I didn't read that link - and having read it, I'm not certain that the person responding to the question was correct.

I'm not familiar with how linux uses a swapfile, the responder seems to feel that the size of the swapfile is fixed and he may or may not be correct, I do know that other operating systems (Windows, OSX) do use variable size swapfiles.

Think of the swapfile as the random access storage medium - it has a constant size, but within it, there are individual blocks that are constantly being randomly written, read, and overwritten - fragmentation is possible.

Consider this - fragmentation occurs in dynamic RAM - i(the DIMMs of your PC) - you think it won't occur in virtual memory (the swapfiles)?

I'm not going to get into a discussion on swap file vs. swap partition.

I got into this on - fragmentation is not an issue on flash memory - which is incorrect, fragmentation will and does occur on any random access storage medium that allows read/write/delete with variable file (or block) sizes - period.