Page 38 of 55 FirstFirst ... 28333435363738394041424348 ... LastLast
Results 741 to 760 of 1081
Like Tree13Likes
  1. #741  
    Quote Originally Posted by graffix31 View Post
    Ok. Clarrification def. Needed On this. i found everything you are talking about in the fstab file just wonderint exactly what i need to do in order to do that you said. am i commenting something out? am i deleting something completely? i sent you a pm and then i found the file so i just need a little more help to finish it up.
    thanks a lot.


    -- Sent from my Palm Pre using Forums
    So you know when the pre use to crash or reboot and some of the apps would act like you just started them for the first time. Well that is because of "data=writeback". What that says is to buffer writes and then write them when you have a chance or forced ,aka proper shutdown. Most of the time those writes are never made permanent because we never turn off our phones until they crash. By removing "data=writeback" the phone will commit writes as soon as they are made solving some of the data loss issues.
    Sprint pre -> Motorola Photon 4G
  2. rksand's Avatar
    Posts
    151 Posts
    Global Posts
    152 Global Posts
    #742  
    Quote Originally Posted by theXfactor2011 View Post
    So you know when the pre use to crash or reboot and some of the apps would act like you just started them for the first time. Well that is because of "data=writeback". What that says is to buffer writes and then write them when you have a chance or forced ,aka proper shutdown. Most of the time those writes are never made permanent because we never turn off our phones until they crash. By removing "data=writeback" the phone will commit writes as soon as they are made solving some of the data loss issues.
    I think the "clarification needed" is exactly what is being removed other that "data=writeback"?

    i.e. just those words from line 3
    Code:
    /dev/mapper/store-var   /var    ext3    noatime,data=writeback  0       0
    I don't want to make changes unless I'm sure.
    Last edited by rksand; 06/28/2011 at 09:29 AM.
  3. #743  
    Quote Originally Posted by rksand View Post
    I think the "clarification needed" is exactly what is being removed other that "data=writeback"?

    i.e. just those words from line 3
    Code:
    /dev/mapper/store-var   /var    ext3    noatime,data=writeback  0       0
    ...or comment out the whole line. I believe the former, but don't want to make changes unless I'm sure.

    First and foremost. Edit your post now and delete that bit about commenting out that line. It's insanity and will make the phone not boot.

    Secondly, this isn't hard. If someone doesn't understand what they are doing, and doesn't even understand how to properly edit a config file on Linux, they shouldn't be. Period. This is really basic stuff... and it's being applied to the file that tells the kernel where all the partitions are. If that file gets messed up in anyway, your machine won't boot. Period.

    When it's said to "Remove the 'data=writeback' part", I don't understand where the problem is. If something that specific is causing confusion, exit nano, close your novaterm session, and forget it was even mentioned.


    M.
  4. snpalavan's Avatar
    Posts
    71 Posts
    Global Posts
    72 Global Posts
    #744  
    I just substituted the default value: instead of "=writeback", i put "=ordered". That way i have more ease of mind that that value is used and less worried that I am deleting things completely.
  5. #745  
    Quote Originally Posted by rksand View Post
    I think the "clarification needed" is exactly what is being removed other that "data=writeback"?

    i.e. just those words from line 3
    Code:
    /dev/mapper/store-var   /var    ext3    noatime,data=writeback  0       0
    I don't want to make changes unless I'm sure.
    that is exactly what i was asking. not really sure why all the hostility. an answer would have been much shorter and easier than all of the unneeded you suck and dont deserve to do it if you dont know how. i will do what you did and change write back to ordered instead of removing it and not knowing exaclty how much to remove from the line. i guess we shouldnt ask for help anymore. im sorry im just shocked with the change in attitude in this thread
    thanks
  6. #746  
    Quote Originally Posted by graffix31 View Post
    an answer would have been much shorter and easier than all of the unneeded you suck and dont deserve to do it if you dont know how.
    What I said was (essentially) the fstab is NOT the place to play around and find out if one can do it. Same goes the the person who PM'ed me about my response.


    M.
  7. #747  
    i understand it isnt the place to play around. i think everything in this thread comes with the same disclaimer. messing up could result in total meltdown and a dr. visit. i have been able to meta dr. using the new test now main scripts and the patch to add the bigger swap partition. i have been playing with sys ctl and compcache size and swap and all that. i appreciate the concern i really do. i believe i was the one who pm'd you as well. i do not fully understand everything but really like learning about it so thank you for all of your input and help so far in this thread.
  8. #748  
    Quote Originally Posted by graffix31 View Post
    that is exactly what i was asking. not really sure why all the hostility. an answer would have been much shorter and easier than all of the unneeded you suck and dont deserve to do it if you dont know how. i will do what you did and change write back to ordered instead of removing it and not knowing exaclty how much to remove from the line. i guess we shouldnt ask for help anymore. im sorry im just shocked with the change in attitude in this thread
    thanks
    No offense but the hostility is needed. Editing config files can brick your phone if not done right, hence why the stressing that if one is not 100% on what to do, they should not do it (I myself learned by making the mistake of leaving Internalz on EOL = Windows). Plus, written instructions always run the risk of being interpreted wrongly, as opposed to someone showing you how so while the instructions are right, the end user may interpret placement of commas, semicolons, spaces, quotes, etc. wrongly, which happen to be important pieces. As clearly said, it was said to remove that tidbit but this was interpreted to mean different things, which no amount of clarification can ensure that this can be interpreted one way. But a good, safe way to do things like this is when in doubt, don't remove it but modify it. For this case, changing the word writeback to ordered. But whenever in doubt, don't remove, and never, ever, comment out lines, especially when they are multiple things going on. Remember, no one is attacking anyone, just being brutally honest to prevent you from having a dead phone (I've also been through this twice, with NuROM on my TX, which you think I'd have learned the first time).

    Personally, I wonder why no one has made a patch yet to do this safely?
  9. rksand's Avatar
    Posts
    151 Posts
    Global Posts
    152 Global Posts
    #749  
    So to be clear for everyone. To minimize the risk of data loss during a hard crash, edit the /etc/fstab file. Change line 3 from:

    Code:
    /dev/mapper/store-var   /var    ext3    noatime,data=writeback  0       0
    to read:

    Code:
    /dev/mapper/store-var   /var    ext3    noatime,  0       0
    or if you prefer:

    Code:
    /dev/mapper/store-var   /var    ext3    noatime,data=ordered  0       0
    The mention in post #735 is rather casual and does not emphasize the risk that has become so apparent later in this thread.
  10. #750  
    So Its been a little more than a week and I haven't rebooted my pre. After angry birds, web browsing, form reading, etc its still smooth. So here were my settings:

    503mb of swap space

    sysctl.conf
    vm.swappiness = 60
    vm.vfs_cache_pressure = 200
    vm.dirty_expire_centisecs = 1500
    vm.dirty_writeback_centisecs = 800
    vm.page-cluster = 3
    vm.dirty_background_ratio = 10
    vm.dirty_ratio = 40
    vm.min_free_kbytes = 2048

    I also made a little change that I have no clue if it does anything at all:
    in /etc/palm/luna.conf I changed UiMainLow=256
    its default value was 512. This setting is under [CpuShare]

    My thought was that these are the values where the system flags low memory modes or starts trying to free space etc. But I could be completely wrong. 512mb is all the newer phones have so they should be in the same boat. Either way I changed it and nothing broke.


    Hows everyone else doing?
    Sprint pre -> Motorola Photon 4G
  11. rksand's Avatar
    Posts
    151 Posts
    Global Posts
    152 Global Posts
    #751  
    I can report with 500mb swap I've tried both compcache on and off and while general use is much better, I'm still finding it bogs down significantly after a 1 1/2 to 2 days use. I will try your changes and report back. I'd love to get a week!
  12. #752  
    Quote Originally Posted by theXfactor2011 View Post
    So Its been a little more than a week and I haven't rebooted my pre. After angry birds, web browsing, form reading, etc its still smooth. So here were my settings:

    503mb of swap space

    sysctl.conf
    vm.swappiness = 60
    vm.vfs_cache_pressure = 200
    vm.dirty_expire_centisecs = 1500
    vm.dirty_writeback_centisecs = 800
    vm.page-cluster = 3
    vm.dirty_background_ratio = 10
    vm.dirty_ratio = 40
    vm.min_free_kbytes = 2048

    I also made a little change that I have no clue if it does anything at all:
    in /etc/palm/luna.conf I changed UiMainLow=256
    its default value was 512. This setting is under [CpuShare]

    My thought was that these are the values where the system flags low memory modes or starts trying to free space etc. But I could be completely wrong. 512mb is all the newer phones have so they should be in the same boat. Either way I changed it and nothing broke.


    Hows everyone else doing?
    2 questions:

    1. Can we confirm that the numbers are measured in MB?
    2. Can we confirm that those are memory related settings? If so, would those values be the points that Luna maxes out to and starts to garbage collect itself?

    For reference, it's also 512 for 1.4.5.

    As to my current status, I'm still using what I previosly posted and it's working great. My swap usage settles at around 120-130 MB. There are occasional stalls (especially when answering a call) but there have been no lock-ups. I did encounter a few random Luna resets but I have no idea what triggered them (swap was 120-140 when I checked 5 minutes before the reset).

    A question to all though. It might be because I'm confusing how ramzswap functions but I noticed that the ramzswap parition never gets above half way. Is this related to my use of disk_size vs. mem_limit in my script? Is the value chosen in the script the original data size or the partition size? Any idea what I'm doing wrong to not be able to utilize the full ramzswap partition? Perhaps I should change the priority level?
  13. #753  
    Hey guys

    I have been messing around with sysctl for a week and rather than updating constantly I thought I would do some tests without releasing the info.

    This setting so far has been awesome for me.

    Code:
    vm.swappiness = 40
    vm.vfs_cache_pressure = 400
    vm.dirty_expire_centisecs = 500
    vm.dirty_writeback_centisecs = 200
    vm.dirty_background_ratio = 5
    vm.dirty_ratio = 3
    vm.min_free_kbytes = 1024
    vm.max_map_count = 32768
    vm.overcommit_memory = 0
    vm.overcommit_ratio = 10
    vm.page-cluster = 6
    vm.zone_reclaim_mode = 2
    kernel.shmax = 268435456
    kernel.shmall = 16777216
    sched_compat_yield = 1
    vm.lowmem_reserve_ratio = 1024
    (I am using 512mb of swap with this and no compcache)
  14. #754  
    Tomorrow I am going to write an entirely new thread detailing every step to get WebOS 2.1 running smoothly on the Pre Minus.

    The different steps will include:

    -Overclocking
    -Modifying sysctl.conf
    -Patches and apps that speed up the Pre
    -Minimal Logging
    -Modifying Luna.conf
    -Making a larger swap file
    -Changing Backing Store to Linux Swap.
    -Disabling Write buffers

    I have also come across some other things that we haven't discussed here yet.
  15. #755  
    Quote Originally Posted by rmausser View Post
    Tomorrow I am going to write an entirely new thread detailing every step to get WebOS 2.1 running smoothly on the Pre Minus.

    The different steps will include:

    -Overclocking
    -Modifying sysctl.conf
    -Patches and apps that speed up the Pre
    -Minimal Logging
    -Modifying Luna.conf
    -Making a larger swap file
    -Changing Backing Store to Linux Swap.
    -Disabling Write buffers

    I have also come across some other things that we haven't discussed here yet.
    I would request that instead of fragmenting all the information into a new thread where only you can edit the post, that you instead add the information to the existing webOS 2.1.0 Upgrade page on the WebOS Internals wiki (or on a new page linked from there).

    Then all of the folks involved in all those things can collaborate in the one place ...

    Thanks,

    -- Rod
    WebOS Internals and Preware Founder and Developer
    You may wish to donate by Paypal to donations @ webos-internals.org if you find our work useful.
    All donations go back into development.
    www.webos-internals.org twitter.com/webosinternals facebook.com/webosinternals
  16. #756  
    That would be the best Rod, you are right.

    I will do that.
  17. #757  
    im assuming of course wherever you put it you will post the link here for us to follow. :-)
  18. #758  
    Using Internals Pro on my Sprint Pre- with 1.4.5, I dont see a file or folder named compcache in etc/event.d

    Is this normal?
    Last edited by NABRIL15; 07/01/2011 at 09:37 AM.
  19. #759  
    are you on 1.4.5 or 2.1?
  20. #760  
    sorry....1.4.5..Sprint Pre-

Posting Permissions