Has anyone else seen this & is there a solution ?
--
========================,,============================================
SUPPORT ___________//___, Philip Webb
ELECTRIC /] [] [] [] [] []| Cities Centre, University of Toronto
TRANSIT `-O----------O---' purslowatchassdotutorontodotca
er... routes? localhost in /etc/hosts? Sounds weird but in the XFree86
days I had not mapped the word localhost to 127 properly, and X behaved
very slow. Just a wild guess though.
Alternatively, reboot to make sure you've unloaded / reloaded modules.
What did you upgrade from? 7.x? 6.x?
--
Iain Buchanan <iaindb at netspace dot net dot au>
When a fly lands on the ceiling, does it do a half roll or a half loop?
Could you expand a little ... (smile) ?
> localhost in /etc/hosts ? Sounds weird, but in the XFree86 days
> I had not mapped the word localhost to 127 properly
> and X behaved very slow. Just a wild guess though.
The uncommented lines in that file are
127.0.0.1 localhost
::1 localhost
Ok, a wild guess (grin).
> Alternatively, reboot to make sure you've unloaded / reloaded modules.
I don't have any modules except Nvidia & I reboot at least daily.
> What did you upgrade from ? 7.x ? 6.x ?
7.3 IIRC
Anyone else see this or have advice ?
Run top and see if something strange is there. No clue what but maybe
there will be a clue there. May try iotop to see if the drive is a
bottle neck. If it is, check to make sure the drives are using DMA or
whatever is fastest for your system. Could even run hdparm -Tt to test
the drives as well.
Just throwing out ideas here.
Dale
:-) :-)
only thinking down the localhost trail again - make sure 127 is routed
through lo.
other than that the disk / cpu does nothing? Run lsmod and see if the
nvidia driver is actually being used? (I have 48 in the used-by column).
Is the general refresh rate and glx type apps ok?
I don't think you need to rebuild nvidia, but did you anyway? You've
probably answered most of these yourself already, but it doesn't hurt to
check...
--
Iain Buchanan <iaindb at netspace dot net dot au>
Time flies like an arrow. Fruit flies like a banana.
> On Tue, 2009-11-10 at 18:19 -0500, Philip Webb wrote:
>> 091111 Iain Buchanan wrote:
>>> On Tue, 2009-11-10 at 16:31 -0500, Philip Webb wrote:
>>>> Since upgrading to Xorg-x11-7.4-r1 , Xterm & Xpdf are very slow
>>>> to open:
>>>> Xterm takes c 6 sec to start & Xpdf also has a noticeable delay.
>>> routes ?
>>
>> Could you expand a little ... (smile) ?
It really sounds like a network timeout issue (I've had response
issues like that with KDE when my ISP is having troubles).
WAGs:
Verify your dns addresses. Maybe reverse their order or even try
opendns.
Verify your network is working normally: Visit a few sites. Ping you
ISP and other sites. Run a network speed test.
"tail -f /var/log/messages" in one window while opening another.
If you have a firewall, check it's log.
Verify that you are using nvidia drivers instead of nv.
Exam Xorg.log for any hints of problems.
HTH,
Roy
I just tested starting Xterm on the desktop where Gkrellm runs.
There is a brief CPU load of 27 % , then 100 % at every start;
at the 1st start, there's a spike in disk access;
1st start was c 6 sec , 2nd c 4 sec . Xpdf is very similar.
Xclock has the same CPU effect, but starts after c 3 sec .
I remerged Xterm & Xclock, ran 'X -configure' & re-installed xorg.conf ,
but there is no change.
> Run lsmod and see if the nvidia driver is actually being used ?
> -- I have 48 in the used-by column.
Yes, it is: used by '26'.
> Is the general refresh rate and glx type apps ok ?
There hasn't been any other similar effect which I've noticed.
> I don't think you need to rebuild nvidia, but did you anyway ?
I last rebuilt Nvidia-drivers after the Xorg-x11 upgrade;
I had to use the testing version 185.18.36-r1 ,
as "stable" 180.60 wouldn't compile with Gentoo-sources 2.6.31-r4 .
Someone else mentioned networking, but I have no local network or firewall
& none of these apps uses the Internet.
The localhost issue in /etc/hosts only affected KDE as far as I
recall. If you are not running KDE as your DE then it would be
something else. Over here xpdf takes less that 1 second to pop up
when started from a terminal in Fluxbox, but I am not running
xorg-x11.
--
Regards,
Mick
What are you running to support graphics for Fluxbox ??
Is this what you mean, or are you talking about the drivers that I use?
==========================================
[I] x11-base/xorg-server
Installed versions: 1.6.3.901-r2(16:38:59 10/06/09)(hal nptl sdl xorg -
debug -dmx -ipv6 -kdrive -minimal -tslib)
Homepage: http://xorg.freedesktop.org/
Description: X.Org X servers
==========================================
--
Regards,
Mick
So you are running Xorg-x11 ! -- sorry for the quibble.
There is a problem somewhere & it mb a small fix for my machine
& it wb nice to track it down if others are not experiencing it.
> There is a problem somewhere & it mb a small fix for my machine
> & it wb nice to track it down if others are not experiencing it.
Try 'strace xterm' in a terminal with long scrollback history, press the
enter key a couple of times when it hangs, and analyze the output at these
positions later. Or send these parts to the list, maybe someone here will
spot what's going on.
Wonko
> The localhost issue in /etc/hosts only affected KDE as far as I
> recall.
Unfortunately not. I've had it a few times with gnome, but not since
the early days... I think it was an X issue, not related to the DE,
afair :)
--
Iain Buchanan <iaindb at netspace dot net dot au>
This door is baroquen, please wiggle Handel.
(If I wiggle Handel, will it wiggle Bach?)
-- Found on a door in the MSU music building
...
> So you are running Xorg-x11 ! -- sorry for the quibble.
For the record xorg-x11 is the meta-package that pulls in lots of stuff,
xorg-server is the minimal ebuild that xorg-x11 would pull in.
Phillip, what version did xorg-server go from/to when you upgraded?
the testing nvidia-drivers look a bit suspicious, is it worth seeing
what the problem was with the stable version?
Why did you pick 185.18.36-r1? Is there a requirement for the 18x
series? The latest "unstable" (but not masked) is 190.42-r2.
--
Iain Buchanan <iaindb at netspace dot net dot au>
"God is a comedian playing to an audience too afraid to laugh."
- Voltaire
I did that earlier just for fun. The output is huge :-)
I see here that while xterm is starting, the output stutters quite a lot and
each time there is a "resource unavailable" message just before. It's trying
to read fd3 which turns out to be the Unix socket in /tmp
If anyone feels like trawling through strace output, I can post what happens
here. xterm is now taking 4 seconds to start whereas before it was almost
instant
--
alan dot mckinnon at gmail dot com
Or better, use the '-o outputfile' option.
>> ... press Enter a couple of times when it hangs
>> and analyze the output at these positions later.
> I did that earlier just for fun. The output is huge :-)
> 2400 lines when I did it with the same result as yours.
> I see here that while xterm is starting, the output stutters quite a lot
> and each time there is a "resource unavailable" message just before.
> It's trying to read fd3 which turns out to be the Unix socket in /tmp .
Yes, there are 3 lock file/dirs there :
root:527 tmp> ls -la
total 5
drwxrwxrwt 7 root root 200 2009-11-17 16:41 .
drwxr-xr-x 23 root root 560 2009-07-04 05:53 ..
drwxrwxrwt 2 root root 88 2009-11-17 12:50 .ICE-unix
drwx------ 3 purslow purslow 192 2009-11-17 16:30 kde-purslow
drwx------ 2 purslow purslow 240 2009-11-17 16:32 ksocket-purslow
-r--r--r-- 1 root purslow 11 2009-11-17 12:47 .X0-lock
drwxrwxrwt 2 root root 72 2009-11-17 12:47 .X11-unix
Are you using Nvidia ? -- someone else mentioned that as a possible cause.
Otherwise, does anyone have further suggestions ?
Yes, nvidia-190.42-r3
I've noticed recently that kdm takes looooooong to start - more than 30
seconds, and xterm is still taking over 3 seconds to load.
I'll try downgrade nvidia tomorrow when I have a chance and see what happens
>
> Otherwise, does anyone have further suggestions ?
>
--
I'm using 185 & couldn't get 180 to compile with Kernel 2.6.31 ,
so it looks as if Nvidia isn't the problem.
> I've noticed recently that kdm takes looooooong to start -
> more than 30 seconds, and xterm is still taking over 3 seconds to load.
I've tried starting it from another Xterm & as root & there's no difference.
I don't use Kdm, but 'startx' from a raw terminal into Fluxbox.
It looks like a bug for the developers of X .
I've tried switching to the nv kernel driver and it made no difference here.
I'm pretty sure it's an X problem. I'd better try reverting X to an older
version.
--
Rgds
Peter
Extensive Googling found 1 useful hint of where the problem might lie:
http://stackoverflow.com/questions/1087580/resource-temporarily-unavailable-in-boost-asio
"'Resource temporarily unavailable' is normally the text description
for EAGAIN, indicating that the operation should be retried ...
It's generally worth looking at the man page
for the underlying libc function; which is recvfrom in this case"
Yes, 'man recvfrom' has rather opaque info which mb relevant ... (sigh)
> Extensive Googling found 1 useful hint of where the problem might lie:
>
>
> http://stackoverflow.com/questions/1087580/resource-temporarily-unavaila
> ble-in-boost-asio
>
> "'Resource temporarily unavailable' is normally the text description
> for EAGAIN, indicating that the operation should be retried ...
> It's generally worth looking at the man page
> for the underlying libc function; which is recvfrom in this case"
>
> Yes, 'man recvfrom' has rather opaque info which mb relevant ... (sigh)
Relevant, perhaps, but not useful - to me, at least. I'll just have to wait
for the next release of X. Thanks anyway.
--
Rgds
Peter