1 |
Mark Knecht posted on Sun, 04 Aug 2013 16:46:22 -0700 as excerpted: |
2 |
|
3 |
> From previous concersations I assume a lot of folks here use KDE. I |
4 |
> run mostly stable and just updated to kde-4.10.5 which included an |
5 |
> update to kdm. It now seems that I'm plagued by a more than 2 minute |
6 |
> delay from entering my password until KDE flashes up the box showing |
7 |
> that KDE is starting up. Once KDE is up and running everything seems |
8 |
> normal. There are no messages in /dev/log/message that identify any |
9 |
> problems. Just a 2+ minute delay before much of anything occurs. |
10 |
> |
11 |
> Here's a small snippet of message showing the basic delay: |
12 |
|
13 |
[Snip log showing, indeed, a two minute delay...] |
14 |
|
15 |
> I Googled around for a while and got curious about whether this was |
16 |
> really a KDE problem so I emerged xfce4. It has the same 2 minute issue |
17 |
> so this appears to me more of a kdm issue as best I can tell right now. |
18 |
> |
19 |
> It's not anything obvious (to me) with networking delays as I can |
20 |
> sit in the console and ping web sites. No delays there. I got back to |
21 |
> the GUI and it's just waiting. I ran iotop in a console and there's |
22 |
> nothing going on. The machine is just hung for a couple of minutes. |
23 |
> |
24 |
> This just started in the last couple of days with this month's KDE |
25 |
> release. |
26 |
|
27 |
While I run kde, I don't run a *dm, preferring to login at the text |
28 |
console and run startx with XSESSION pointed at kde. Additionally, I |
29 |
have USE=-policykit and USE=-consolekit (policykit was showing up in the |
30 |
log before the delay) set and don't have either one even on my system at |
31 |
all. Similarly USE=-udisks and USE=-upower (those showed up after the |
32 |
delay), so I don't have those to worry about either. So I have no direct |
33 |
help for you from that angle. |
34 |
|
35 |
However, what you report sounds very much like a timeout problem, |
36 |
something expected to already be running not being there, especially |
37 |
since both xfce and kde have the same two-minute delay and there's no I/O |
38 |
activity going on. |
39 |
|
40 |
You mentioned running iotop in a text console and it showing no i/o to |
41 |
speak of. What about CPU activity as shown by regular top/htop , and/or |
42 |
load as shown by uptime? If it's a timeout issue as I expect, that won't |
43 |
show any significant activity either. |
44 |
|
45 |
What I think is happening is that the system's waiting for a response, |
46 |
doing nothing but waiting for two minutes, until some timeout, after |
47 |
which it gives up and either starts it on its own or does without, |
48 |
depending on what it is that's missing. |
49 |
|
50 |
I don't know for sure what that might be, but I'd suggest checking to see |
51 |
if you have a system dbus running (dbus initscript), and if not, start |
52 |
that before you try a kde/xfce login, and see if your delay disappears. |
53 |
|
54 |
Another thing to check, since upower was the first thing logged after the |
55 |
delay, is your power profiles, and/or especially if you're on a |
56 |
wall-powered machine anyway, try setting USE=-upower and doing an |
57 |
emerge --newuse @world, and see if that helps. (It's quite possible that |
58 |
upower/powerdevil are entirely innocent and have nothing to do with the |
59 |
problem; that it just happens that they're the first thing logged after |
60 |
the timeout, but it never hurts to check that, just to be sure.) |
61 |
|
62 |
|
63 |
When you do figure out the problem, please post what it was, as I'm |
64 |
curious, now. |
65 |
|
66 |
-- |
67 |
Duncan - List replies preferred. No HTML msgs. |
68 |
"Every nonfree program has a lord, a master -- |
69 |
and if you use the program, he is your master." Richard Stallman |