ltib
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Ltib] Problem using LTIB


From: Stuart Hughes
Subject: Re: [Ltib] Problem using LTIB
Date: Thu, 14 Oct 2010 19:36:05 +0100
User-agent: Thunderbird 2.0.0.16 (X11/20080707)

Hi Daniel,

I just tried here (at home) and it bobbed around 36 K Bytes per second,
which is slow.  I'll need to investigate, can't promise when I can get
to this though as I'm maxed out.

Regards, Stuart

Daniel Brown wrote:
> I'm in California.
> 
> -----Original Message-----
> From: Stuart Hughes [mailto:address@hidden 
> Sent: Thursday, October 14, 2010 4:13 AM
> To: Aaron Wegner
> Cc: address@hidden; Daniel Brown
> Subject: Re: [Ltib] Problem using LTIB
> 
> Hi Guys,
> 
> I'll try downloading that test case file as soon as I get some time and
> report back.
> 
> It would be interesting to hear from anyone that has recently been
> download and gets good bandwidth.
> 
> What's I'd like to figure is whether there's any pattern to this.
> 
> BTW: what's you're location? I'm in the UK.  The server is in California
> AFAIK.
> 
> Regards, Stuart
> 
> Aaron Wegner wrote:
>> Stuart,
>>
>> I tried again just now for the fun of it.  I'm using the file
>> linux-2.6.13.tar.gz as a test case.  Both sites start out fairly well,
>> around 70 KB/sec, but bitshrine.org drops very quickly to 1 KB/sec,
> then
>> after a while bounces back up to 100 KB/sec, then drops again quickly
> to 1
>> KB/sec.  I'm estimating an average of 10 KB/sec or less.  Firefox
> seems
>> capable of completing the download this year, while wget often fails. 
>> With kernel.org I'm floating between 40 KB/sec and 460 KB/sec.  Appear
> to
>> be averaging 120 KB/sec.  It seems like bitshrine.org or globat.com
> might
>> not play nice with machines if it perceives a lower download rate on
> the
>> opposite end.  The result is I am basically not able to download from
> the
>> gpp.  In the past it has worked here.  Maybe it works well if the
> network
>> is unoccupied.  Would be nice it it worked "better" when the network
> is
>> occupied.  Seems like if you have a good network life is good and if
> you
>> don't then you're punished doubly!  I'm not doubting it's related to
>> something in our network, but of all the sites I've been to only the
> gpp
>> is prohibitively slow, so it might be something on that end as well.
> I
>> used to download from work without an issue and have downloaded in the
>> past from home without an issue.  The whole time I've been typing this
>> email I'm at 11% on my kernel.  The toolchains are even bigger.
>>
>> Aaron
>>
>>> Hi Aaron,
>>>
>>> All I can say is that when I nearly always get good bandwidth (a few
>>> hundred Kb/s).  I have regularly had ltib download toolchains at good
>>> download rates.
>>>
>>> The most likely cause is related to your network infrastructure
>>> (firewalls etc).
>>>
>>> The ISP for bitshrine is globat.com, most of the time they're very
> good.
>>>  However if you can see anything incorrect, it might be worth
> contacting
>>> them and letting them know.
>>>
>>> BTW: can you try from another physical location.
>>>
>>> Regards, Stuart
>>>
>>> Aaron Wegner wrote:
>>>> Guys, I have encountered something similar as well.  I took a guess
> that
>>>> it might be ipv6 related, so I disabled that, but really no
> significant
>>>> improvement.  I guessed that because about 2 months ago I tried
>>>> debugging
>>>> the same issue and it seemed like some of the Windows machines here
> had
>>>> more luck with the download rates than my Linux machine was having.
> Now
>>>> I
>>>> can't reproduce that behavior.  Today I've tried Windows, Debian
> Linux,
>>>> and Fedora Linux.
>>>>
>>>> I don't know how the handshakes work, but it seems like periodically
> the
>>>> transfer rate is renegotiated, and always to a very low rate.  Now
> I'm
>>>> achieving 5 KB/sec from bitshrine.org.  Other sites work better for
> me,
>>>> about 300 KB/sec from kernel.org.  I have resorted to downloading my
>>>> .tar.gz files from other sites on the Internet, but the biggest
> problem
>>>> for me is the toolchain, which I would like to get from
> bitshrine.org,
>>>> as
>>>> I don't really know where else to look for the exact same files.
> 100 MB
>>>> at 5 KB/sec means I'll have it next week.
>>>>
>>>> Is it possible that bitshrine.org does not allow renegotiation of
>>>> download
>>>> rate often enough?  I don't know enough about it.
>>>>
>>>> Aaron
>>>>
>>>>> Hi Daniel,
>>>>>
>>>>> Normally the bandwidth is pretty good from that ISP, so it's worth
>>>>> checking your network.
>>>>>
>>>>> Glad you found the problem.  Ideally LTIB could be improved to look
> to
>>>>> see if wget times out and report that, rather than stumbling on and
>>>>> failing the md5sum check.  At least there is a check though so it
>>>>> prevents untaring *rap.
>>>>>
>>>>> Regards, Stuart
>>>>>
>>>>> Daniel Brown wrote:
>>>>>> Stuart,
>>>>>>
>>>>>> Actually, I finally discovered the problem. The download wasn't
>>>>>> finishing
>>>>>> due either to my network or to the server. The download transfer
> rates
>>>>>> were all over the place but at their worst were 1 KB/sec and the
>>>>>> connection
>>>>>> was timing out. I guess the wgets program doesn't catch or send an
>>>>>> appropriate
>>>>>> signal on connection termination but the script, thinking the file
> was
>>>>>> downloaded, probably did an MD5 checksum on a partially downloaded
>>>>>> file.
>>>>>> I edited the wgets options to remove the non-verbose mode and
> improve
>>>>>> the timeout of only 12 seconds. This did better but for some
> reason
>>>>>> the
>>>>>> connections still timed out. There is something wrong with our
> network
>>>>>> which
>>>>>> I am having IT look into. After several tries I was finally able
> to
>>>>>> get
>>>>>> LTIB to download all the necessary RPMs and finish the compile.
>>>>>>
>>>>>> Thanks for the help.
>>>>>> Daniel
>>>>>>
>>>>>> -----Original Message-----
>>>>>> From: Stuart Hughes [mailto:address@hidden
>>>>>> Sent: Saturday, August 21, 2010 3:53 AM
>>>>>> To: Daniel Brown
>>>>>> Cc: address@hidden; address@hidden
>>>>>> Subject: Re: [Ltib] Problem using LTIB
>>>>>>
>>>>>> Hi Daniel,
>>>>>>
>>>>>> That is just for installation so you can run /mnt/ltib/install.
> Make
>>>>>> when you run the installer you are in some area with local disk
>>>>>>
>>>>>> Also, in the output I see the warning:
>>>>>>
>>>>>>>     Ideally the rpm database should not be located in an NFS
> mounted
>>>>>>>     filesystem.
>>>>>>>
>>>>>>>     On some systems this may cause cause problems due to
> filesystem
>>>>>> locking
>>>>>>
>>>>>> If you can avoid running on an NFS area, you'll have less problems
> as
>>>>>> your IT department can setup all sorts of restrictions.
>>>>>>
>>>>>> Finally the message:
>>>>>>>     ERROR: md5sum mismatch, re-naming
>>>>>>>     /opt/ltib/pkgs/tc-fsl-x86lnx-armeabi-nptl-4.1.2-3.i386.rpm to
>>>>>>>
> /opt/ltib/pkgs/tc-fsl-x86lnx-armeabi-nptl-4.1.2-3.i386.rpm.bad,
>>>>>>>     please re-try
>>>>>> This shows there was some kind of download error.  Can you take a
> look
>>>>>> at the .bad file (run file and see what it is, check the size).
> The
>>>>>> idea is to see if the http server returned an error (which will be
> in
>>>>>> the .bad file).  One possibility is that you need to set up a web
>>>>>> proxy
>>>>>> (does your browser use one?).  The other possibility is that the
>>>>>> download aborted for lack of disk space or some other reason (IT
> quota
>>>>>> policy etc?).
>>>>>>
>>>>>> Regards, Stuart
>>>>>>
>>>>>>
>>>>>>
>>>>>> Daniel Brown wrote:
>>>>>>> Hi Peter,
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> I do have the following in my list of mounted devices:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> /home/dbrown/Freescale/MPC8313E_RDB_K26_20081226-LTIB.iso on
>>>>>>> /mnt/ltib
>>>>>>> type iso9660 (rw,loop=/dev/loop0)
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> This was for the Freescale LTIB distribution that I downloaded
>>>>>>> before.
>>>>>>> Is it possible that this is the problem?
>>>>>>>
>>>>>>>
>>>>>>> Thanks.
>>>>>>>
>>>>>>> Daniel
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> *From:* Peter Barada [mailto:address@hidden
>>>>>>> *Sent:* Friday, August 20, 2010 11:07 AM
>>>>>>> *To:* Daniel Brown
>>>>>>> *Cc:* Jorge A. Castro; address@hidden
>>>>>>> *Subject:* RE: [Ltib] Problem using LTIB
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On Fri, 2010-08-20 at 10:55 -0700, Daniel Brown wrote:
>>>>>>>
>>>>>>> Jorge,
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> I am using Redhat Enterprise Workstation. Is this dist okay?
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> My network connection is fine on this machine. I have checked.
> But
>>>>>>> how
>>>>>>> can I check connection to the webite for
>>>>>>>
>>>>>>> getting the tc-fsl package?
>>>>>>>
>>>>>>>
>>>>>>> in the get_remote() function in bin/Ltibutils.pm, if you add a
> "set
>>>>>>> -x"
>>>>>>> to the shell commands that call wget, thereby seingo the wget
> command
>>>>>>> used to fetch the file from the GPP(note the bolded "set -x" in
> the
>>>>>>> following):
>>>>>>>
>>>>>>> sub get_remote
>>>>>>> {
>>>>>>>     my($dest, $pxys, $wget_opts, $pxmode, $url) = @_;
>>>>>>>     return system_nb(<<TXT) == 0;
>>>>>>> *set -x*
>>>>>>> cd $dest
>>>>>>> $pxys wget $wget_opts --proxy=$pxmode $url 2>&1
>>>>>>> cd - >/dev/null
>>>>>>> TXT
>>>>>>> }
>>>>>>>
>>>>>>> >From what you've sent, it looks like you wee able to downloaded
> the
>>>>>>> file, its a mismatch of the md5sum that's causing the renaming.
> I'm
>>>>>>> guessing that you have an LTIB .iso image - are the line endings
>>>>>>> incorrect (i.e. finding DOS line endings (CRLF) instead of Unix
> line
>>>>>>> endings (LF only))?
>>>>>>>
>>>>>>> If a mismatch is found, LTIB moves the filename (by appending
> ".bad"
>>>>>>> to
>>>>>>> it), so the next run of LTIB should pull it down again, and
> hopefully
>>>>>>> it
>>>>>>> will match.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Thanks.
>>>>>>>
>>>>>>> Daniel
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     *From:address@hidden
>>>>>>>     [mailto:address@hidden *On
>>>>>>> Behalf
>>>>>>>     Of *Jorge A. Castro
>>>>>>>     *Sent:* Friday, August 20, 2010 9:31 AM
>>>>>>>     *To:* address@hidden
>>>>>>>     *Subject:* Re: [Ltib] Problem using LTIB
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     Hi Daniel,
>>>>>>>
>>>>>>>     It seems like LTIB it's not getting
>>>>>>>     tc-fsl-x86lnx-armeabi-nptl-4.1.2-3.i386.rpm, because it shows
>>>>>>>
>>>>>>>     Can't get: tc-fsl-x86lnx-armeabi-nptl-4.1.2-3.i386.rpm at
> ./ltib
>>>>>>>     line 2628.
>>>>>>>
>>>>>>>     You may check your connection  or the availability of the
>>>>>>> package.
>>>>>>>     Bytheway, what are you using in you host machine? (what
> GNU/Linux
>>>>>>>     flavor distribution?)
>>>>>>>
>>>>>>>     Regards,
>>>>>>>
>>>>>>>     Jorge Castro
>>>>>>>
>>>>>>>
>>>>>>>     On 08/20/2010 10:16 AM, Daniel Brown wrote:
>>>>>>>
>>>>>>>     Hello,
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     I downloaded and having tried using ltib to build a
> configuration
>>>>>>>     for the LPC3250 target.
>>>>>>>
>>>>>>>     I get the following transcript after the gui exits. Can
> anyone
>>>>>>> help?
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     -------
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     [bash-3.2:dbrown:ltib]$ ./ltib
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     Installing host support packages.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     This only needs to be done once per host, but may take up to
>>>>>>>
>>>>>>>     an hour to complete ...
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     If an error occurs, a log file with the full output may be
> found
>>>>>>> in:
>>>>>>>
>>>>>>>     /home/dbrown/ltib/host_config.log
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     *** End of configuration.
>>>>>>>
>>>>>>>     *** Check the top-level Makefile for additional configuration
>>>>>>> options.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     + '[' -n '' ']'
>>>>>>>
>>>>>>>     + '[' '!' -f config/platform/ea3250/.config -a -f
>>>>>>>     config/platform/ea3250/defconfig ']'
>>>>>>>
>>>>>>>     + cp config/platform/ea3250/defconfig
>>>>>>> config/platform/ea3250/.config
>>>>>>>
>>>>>>>     + '[' -f /home/dbrown/ltib/config/.config ']'
>>>>>>>
>>>>>>>     + '[' -n '' ']'
>>>>>>>
>>>>>>>     + cd config/platform/ea3250
>>>>>>>
>>>>>>>     + mconf main.lkc .config
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     *** End of configuration.
>>>>>>>
>>>>>>>     *** Check the top-level Makefile for additional configuration
>>>>>>> options.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     + cp .config defconfig.dev
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     Ideally the rpm database should not be located in an NFS
> mounted
>>>>>>>     filesystem.
>>>>>>>
>>>>>>>     On some systems this may cause cause problems due to
> filesystem
>>>>>>> locking
>>>>>>>
>>>>>>>     and this application.  If you have this problem, you'll see
> error
>>>>>>>     messages like: "error: cannot get exclusive lock on
>>>>>>> ..../Packages"
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     Press <enter to continue>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     Installing: tc-fsl-x86lnx-armeabi-nptl-4.1.2-3.i386.rpm
>>>>>>>
>>>>>>>     Testing network connectivity for gpp
>>>>>>>
>>>>>>>     OK GPP: is available
>>>>>>>
>>>>>>>     Try tc-fsl-x86lnx-armeabi-nptl-4.1.2-3.i386.rpm from the GPP
>>>>>>>
>>>>>>>     ERROR: md5sum mismatch, re-naming
>>>>>>>     /opt/ltib/pkgs/tc-fsl-x86lnx-armeabi-nptl-4.1.2-3.i386.rpm to
>>>>>>>
> /opt/ltib/pkgs/tc-fsl-x86lnx-armeabi-nptl-4.1.2-3.i386.rpm.bad,
>>>>>>>     please re-try
>>>>>>>
>>>>>>>     Can't get: tc-fsl-x86lnx-armeabi-nptl-4.1.2-3.i386.rpm at
> ./ltib
>>>>>>>     line 2628.
>>>>>>>
>>>>>>>     Died at ./ltib line 2628.
>>>>>>>
>>>>>>>     traceback:
>>>>>>>
>>>>>>>      main::check_toolchain_setup:2628
>>>>>>>
>>>>>>>       main::pre_build_checks:1486
>>>>>>>
>>>>>>>        main:564
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     Started: Fri Aug 20 07:20:07 2010
>>>>>>>
>>>>>>>     Ended:   Fri Aug 20 07:43:11 2010
>>>>>>>
>>>>>>>     Elapsed: 1384 seconds
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     VERSION          : 10.1.1
>>>>>>>
>>>>>>>     CVS_VERSION      : $Revision: 1.70 $ (Savannah)
>>>>>>>
>>>>>>>     PLATFORM         : ea3250
>>>>>>>
>>>>>>>     GNUTARCH         : arm
>>>>>>>
>>>>>>>     TOOLCHAIN        :
> tc-fsl-x86lnx-armeabi-nptl-4.1.2-3.i386.rpm
>>>>>>>     TOOLCHAIN_CFLAGS : -O2 -fsigned-char -mfloat-abi=softfp
> -mfpu=vfp
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     Build Failed
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     Exiting on error or interrupt
>>>>>>>
>>>>>>>     [bash-3.2:dbrown:ltib]$
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     _______________________________________________
>>>>>>>
>>>>>>>     LTIB home page: http://ltib.org
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     Ltib mailing list
>>>>>>>
>>>>>>>     address@hidden <mailto:address@hidden>
>>>>>>>
>>>>>>>     http://lists.nongnu.org/mailman/listinfo/ltib
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     --
>>>>>>>
>>>>>>>     Jorge A. Castro
>>>>>>>
>>>>>>>     Canam-Technology, Inc.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     _______________________________________________
>>>>>>>
>>>>>>>     LTIB home page: http://ltib.org
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>     Ltib mailing list
>>>>>>>
>>>>>>>     address@hidden <mailto:address@hidden>
>>>>>>>
>>>>>>>     http://lists.nongnu.org/mailman/listinfo/ltib
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
> ------------------------------------------------------------------------
>>>>>>> _______________________________________________
>>>>>>> LTIB home page: http://ltib.org
>>>>>>>
>>>>>>> Ltib mailing list
>>>>>>> address@hidden
>>>>>>> http://lists.nongnu.org/mailman/listinfo/ltib
>>>>> _______________________________________________
>>>>> LTIB home page: http://ltib.org
>>>>>
>>>>> Ltib mailing list
>>>>> address@hidden
>>>>> http://lists.nongnu.org/mailman/listinfo/ltib
>>>>>
>>>>
>>>> _______________________________________________
>>>> LTIB home page: http://ltib.org
>>>>
>>>> Ltib mailing list
>>>> address@hidden
>>>> http://lists.nongnu.org/mailman/listinfo/ltib
>>>>
>>
>>
>> _______________________________________________
>> LTIB home page: http://ltib.org
>>
>> Ltib mailing list
>> address@hidden
>> http://lists.nongnu.org/mailman/listinfo/ltib
>>
> 



reply via email to

[Prev in Thread] Current Thread [Next in Thread]