Bug 17995 - Booting x86-64 takes double the time of an i586
Summary: Booting x86-64 takes double the time of an i586
Status: RESOLVED FIXED
Alias: None
Product: Mageia
Classification: Unclassified
Component: RPM Packages (show other bugs)
Version: Cauldron
Hardware: All Linux
Priority: Normal normal
Target Milestone: ---
Assignee: Mageia Bug Squad
QA Contact:
URL:
Whiteboard: NEEDINFO
Keywords:
Depends on:
Blocks:
 
Reported: 2016-03-14 17:53 CET by Herman Viaene
Modified: 2017-01-07 13:28 CET (History)
3 users (show)

See Also:
Source RPM:
CVE:
Status comment:


Attachments
journal file (242.11 KB, text/plain)
2016-03-18 10:22 CET, Herman Viaene
Details
Journal boot RC (505.06 KB, text/plain)
2016-07-19 15:35 CEST, Herman Viaene
Details
journal of rc dd 22 july (262.11 KB, text/plain)
2016-07-27 22:41 CEST, Herman Viaene
Details

Description Herman Viaene 2016-03-14 17:53:51 CET
Description of problem:
Booting MGA6 on a recent Lenovo B50 (quadcore, 8 Gb memory) takes double the time than an old Acer D620 (single core , 1 Gb memory)


Version-Release number of selected component (if applicable):


How reproducible:


Steps to Reproduce:
1.
2.
3.
Herman Viaene 2016-03-14 17:54:04 CET

Keywords: (none) => 6dev1

Comment 1 Marja Van Waes 2016-03-18 09:36:41 CET
Maybe related to bug 17574

For the slow booting system, please attach journal.txt that is the result of running, as root:

   journalctl -ab > journal.txt

CC: (none) => marja11

Marja Van Waes 2016-03-18 09:37:17 CET

Keywords: (none) => NEEDINFO

Comment 2 Herman Viaene 2016-03-18 10:22:54 CET
Created attachment 7581 [details]
journal file
Comment 3 Marja Van Waes 2016-03-25 11:13:46 CET
Btw, what is "double" the time?

I see there's 5 minutes between 
mrt 18 10:11:08 mach5.hviaene.thuis sddm-greeter[3025]: Message received from daemon: HostName

and

mrt 18 10:16:46 mach5.hviaene.thuis sddm-greeter[3025]: Reading from "/usr/share/xsessions/01plasma.desktop"

But I suppose that you had walked away from the system, or does it really take 7 minutes from starting the machine till seeing the login screen?


Btw, there are 29 messages about a device having the /sda/sda4 sysfs path, but at the same time having /sda/sda7, or /sda/sda6, or /sda/sda5 sysfs path

mrt 18 10:09:27 mach5.hviaene.thuis systemd[1]: dev-disk-by\x2dpartlabel-Basic\x5cx20data\x5cx20partition.device: Dev dev-disk-by\x2dpartlabel-Basic\x5cx20data\x5cx20partition.device appeared twice with different sysfs paths /sys/devices/pci0000:00/0000:00:1f.2/ata1/host0/target0:0:0/0:0:0:0/block/sda/sda4 and /sys/devices/pci0000:00/0000:00:1f.2/ata1/host0/target0:0:0/0:0:0:0/block/sda/sda5

(See also https://github.com/systemd/systemd/issues/2677 ... another user, whose report was closed as a duplicate, said his system boots up very slowly https://github.com/systemd/systemd/issues/2705 )

But I don't know whether this can cause slow booting, CC'ing Colin

CC: (none) => mageia

Comment 4 Herman Viaene 2016-03-25 11:40:03 CET
I did not walk away from the system, it happens without any intervantion  being possible.
But I noted something different. I routinely install nntp at installation time (when the system asks for the time zone). In this journal there are items on time zone which I would not expect if this is handled correctly.
I checked MCC and found there that the clock synchronization was not set. I ticked it on, and found that it that not ask for the nntp package to install. This i would expect if I had forgotten to tick it on at installation time. I set it on in MCC and since then boot time has improved quite a bit, but not to the same level as M5.
In cas you might ask: the journal I attached to bug 18028 today is of the same system with currently nntp configured OK.
Oh, about the disk partitions: this laptop has 4 running systems: Win 10 (having 4 partitions of its own), an M5 "production" having a / and a /home partition, an M5 for testing and this M6 for testing , so it might look  a bit complicated.
So AFAICS, the slow boot has to do with time zone settings, but that shouldn't have such effect I presume???
Comment 5 Marja Van Waes 2016-03-25 12:14:29 CET
(In reply to Marja van Waes from comment #3)
> Btw, what is "double" the time?
> 
> I see there's 5 minutes between 
> mrt 18 10:11:08 mach5.hviaene.thuis sddm-greeter[3025]: Message received
> from daemon: HostName
> 
> and
> 
> mrt 18 10:16:46 mach5.hviaene.thuis sddm-greeter[3025]: Reading from
> "/usr/share/xsessions/01plasma.desktop"
> 
> But I suppose that you had walked away from the system, or does it really
> take 7 minutes from starting the machine till seeing the login screen?
> 
> 

(In reply to Herman Viaene from comment #4)
> I did not walk away from the system, it happens without any intervantion 
> being possible.
> But I noted something different. I routinely install nntp at installation
> time (when the system asks for the time zone). In this journal there are
> items on time zone which I would not expect if this is handled correctly.

you're right, the 5 minute hang between the sddm-greeter lines that I pasted happens here:

mrt 18 10:11:28 mach5.hviaene.thuis systemd-timesyncd[783]: Synchronized to time server 216.239.34.15:123 (time2.google.com).
mrt 18 10:16:44 mach5.hviaene.thuis chronyd[1010]: Can't synchronise: no majority

> I checked MCC and found there that the clock synchronization was not set. I
> ticked it on, and found that it that not ask for the nntp package to
> install. This i would expect if I had forgotten to tick it on at
> installation time. I set it on in MCC and since then boot time has improved
> quite a bit, but not to the same level as M5.

Ben hit this, too, see bug 17563 which is as good as certain a duplicate of bug 17091

<snip>

> So AFAICS, the slow boot has to do with time zone settings, but that
> shouldn't have such effect I presume???

Maybe sddm can't handle a problem there?

CC'ing neoclust

CC: (none) => mageia

Comment 6 Marja Van Waes 2016-03-25 12:15:21 CET
@ neoclust

What I meant is: can't sddm-greeter just continue, regardless of time problems?
Comment 7 Marja Van Waes 2016-03-28 21:58:13 CEST
(In reply to Marja van Waes from comment #6)
> @ neoclust
> 
> What I meant is: can't sddm-greeter just continue, regardless of time
> problems?

assigning to you, because, other than maybe closing as dup of bug 17091, I wouldn't know what to do with it :-þ

Keywords: NEEDINFO => (none)
Assignee: bugsquad => mageia

Comment 8 Herman Viaene 2016-07-19 15:35:42 CEST
Created attachment 8209 [details]
Journal boot RC
Comment 9 Herman Viaene 2016-07-19 15:38:46 CEST
This problem disappeared in the subsequent sta1 ISO's, but is back in RC, but not so extreme. You can see in the journalrc.txt that there are gaps at 14:34:31 and 14:34:56
Comment 10 Herman Viaene 2016-07-27 22:39:21 CEST
Again in RC of 22 July. Journal boot.txt uploaded,showing jumps at 21:35:31 to 21:35:49 and immediately to 21:36:29.
Comment 11 Herman Viaene 2016-07-27 22:41:03 CEST
Created attachment 8266 [details]
journal of rc dd 22 july
Comment 12 Marja Van Waes 2016-07-28 07:42:00 CEST
(In reply to Herman Viaene from comment #11)
> Created attachment 8266 [details]
> journal of rc dd 22 july

Thanks :-)

At first sight, I see two sddm segfaults.

There is nothing about systemd-timesyncd.

Do you mind also attaching boot.svg, that is the result of running:

   systemd-analyze plot > boot.svg

(Just to rule out that there's something that gives more delay than the sddm segfaults)
Comment 13 Marja Van Waes 2016-07-28 07:47:24 CEST
(In reply to Herman Viaene from comment #10)
> Again in RC of 22 July. Journal boot.txt uploaded,showing jumps at 21:35:31
> to 21:35:49 and immediately to 21:36:29.

Sorry, I missed this comment... so there's indeed a time consumer before sddm was started for the first time!
Comment 14 Marja Van Waes 2016-07-28 08:03:33 CEST
Those sddm segfaults are bug 17653, btw.

Removing neoclust as assignee, because the time jumps aren't related.

Assignee: mageia => bugsquad

Comment 15 Marja Van Waes 2017-01-07 11:53:49 CET
@ Herman

Is this bug still valid?

If so, do you mind replacing attaching a fresh "journalctl -ab" output?
Marja Van Waes 2017-01-07 11:54:02 CET

Whiteboard: (none) => NEEDINFO

Comment 16 Herman Viaene 2017-01-07 13:28:51 CET
No, I haven't noticed such timings anymore with the latest sta's.

Status: NEW => RESOLVED
Resolution: (none) => FIXED


Note You need to log in before you can comment on or make changes to this bug.