headers
John Hasler | 30 Aug 15:41
Favicon

Bug#497113: chrony: /dev/rtc renamed to /dev/rtc0 with linux-image-2.6-*/2.6.26+15

I will work on this this weekend.
--

-- 
John Hasler
Permalink | Reply |
headers
John Hasler | 1 Sep 22:08

Bug#497113: chrony: /dev/rtc renamed to /dev/rtc0 with linux-image-2.6-*/2.6.26+15

Note that rtc is disabled by default.  Only people who have enabled it by
hand will run into this problem.
--

-- 
John Hasler 
john <at> dhh.gt.org
Elmwood, WI USA
Permalink | Reply |
headers
John Hasler | 1 Sep 22:50

Bug#497113: chrony: /dev/rtc renamed to /dev/rtc0 with linux-image-2.6-*/2.6.26+15

According to the udev changelog an rtc symlink is created when the new
drivers are used.  This symlink apparently was added with version 0.116.
--

-- 
John Hasler 
john <at> dhh.gt.org
Elmwood, WI USA
Permalink | Reply |
headers
John Hasler | 1 Sep 23:36

Bug#497113: chrony: /dev/rtc renamed to /dev/rtc0 with linux-image-2.6-*/2.6.26+15

Chrony shouldn't hang when it can't find the rtc: it should just exit.
--

-- 
John Hasler 
john <at> dhh.gt.org
Elmwood, WI USA
Permalink | Reply |
headers
Francesco Poli | 2 Sep 00:03
Favicon

Bug#497113: chrony: /dev/rtc renamed to /dev/rtc0 with linux-image-2.6-*/2.6.26+15

On Mon, 01 Sep 2008 16:36:51 -0500 John Hasler wrote:

> Chrony shouldn't hang when it can't find the rtc: it should just exit.

On the box where I encountered the problem, it seemed to never start.
Since that box is a production one, I could not wait for a long time in
front of a reboot: after some minutes, I rebooted again and all the
rest (as explained in my initial message).

I agree that it *should not* hang: maybe *this* is the bug...

--

-- 
 http://frx.netsons.org/doc/index.html#nanodocs
 The nano-document series is here!
..................................................... Francesco Poli .
 GnuPG key fpr == C979 F34B 27CE 5CD8 DC12  31B5 78F4 279B DD6D FCF4
Permalink | Reply |
headers
John Hasler | 2 Sep 02:47

Bug#497113: chrony: /dev/rtc renamed to /dev/rtc0 with linux-image-2.6-*/2.6.26+15

I wrote:
> Chrony shouldn't hang when it can't find the rtc: it should just exit.

And on my machines it does.

The initscript may hang for a few minutes when netstat can't get the
routing tables, though.  Perhaps that is what you are seeing?
--

-- 
John Hasler 
john <at> dhh.gt.org
Elmwood, WI USA
Permalink | Reply |
headers
Francesco Poli | 2 Sep 19:42
Favicon

Bug#497113: chrony: /dev/rtc renamed to /dev/rtc0 with linux-image-2.6-*/2.6.26+15

On Mon, 01 Sep 2008 19:47:32 -0500 John Hasler wrote:

> I wrote:
> > Chrony shouldn't hang when it can't find the rtc: it should just exit.
> 
> And on my machines it does.

Awkward, I've just tried to discard the rtcdevice directive
from /etc/chrony/chrony.conf
After

  # /etc/init.d/chrony stop
  # /etc/init.d/chrony start

chronyd started and went on running, even though it logged
the following error messages to /var/log/daemon.log:

  chronyd[16196]: Could not open /dev/rtc, No such file or directory
  chronyd[16196]: Real time clock not supported on this operating system

When I first rebooted the box with Linux kernel 2.6.26,
the above error messages were logged, but then chronyd did nothing
else for about 95 s ...
Maybe I was just too impatient?

> 
> The initscript may hang for a few minutes when netstat can't get the
> routing tables, though.  Perhaps that is what you are seeing?

Frankly speaking, I don't know...
(Continue reading)

Permalink | Reply |
headers
John Hasler | 2 Sep 20:34

Bug#497113: chrony: /dev/rtc renamed to /dev/rtc0 with linux-image-2.6-*/2.6.26+15

Francesco Poli writes:
> ...I've just tried to discard the rtcdevice directive...

"rtcdevice" merely tells chronyd that the rtc is in a nonstandard place.
In its absence /dev/rtc is assumed.  To stop chronyd from messing with the
rtc comment out the "rtcfile" directive (I agree that this is confusing).

> When I first rebooted the box with Linux kernel 2.6.26, the above error
> messages were logged, but then chronyd did nothing else for about 95 s
> ...  Maybe I was just too impatient?

The "netstat" call in the initscript can hang longer than that (yes, this
is a bug).
--

-- 
John Hasler 
john <at> dhh.gt.org
Elmwood, WI USA
Permalink | Reply |
headers
Francesco Poli | 2 Sep 22:13
Favicon

Bug#497113: chrony: /dev/rtc renamed to /dev/rtc0 with linux-image-2.6-*/2.6.26+15

On Tue, 02 Sep 2008 13:34:18 -0500 John Hasler wrote:

> Francesco Poli writes:
> > ...I've just tried to discard the rtcdevice directive...
> 
> "rtcdevice" merely tells chronyd that the rtc is in a nonstandard place.

Yes, I am aware of that.
It's precisely why I inserted that directive
into /etc/chrony/chrony.conf: I currently have no /dev/rtc,
but /dev/rtc0 instead.

And this will hold until I upgrade udev.

> In its absence /dev/rtc is assumed.

Yes, that's why chronyd logs error messages in its absence: my system
currently lacks /dev/rtc.

> To stop chronyd from messing with the
> rtc comment out the "rtcfile" directive (I agree that this is confusing).

But this would entirely disable the use of RTC, wouldn't it?

> 
> > When I first rebooted the box with Linux kernel 2.6.26, the above error
> > messages were logged, but then chronyd did nothing else for about 95 s
> > ...  Maybe I was just too impatient?
> 
> The "netstat" call in the initscript can hang longer than that (yes, this
(Continue reading)

Permalink | Reply |
headers
Francesco Poli | 1 Sep 23:56
Favicon

Bug#497113: chrony: /dev/rtc renamed to /dev/rtc0 with linux-image-2.6-*/2.6.26+15

On Mon, 01 Sep 2008 15:50:28 -0500 John Hasler wrote:

> According to the udev changelog an rtc symlink is created when the new
> drivers are used.  This symlink apparently was added with version 0.116.

Good, this should save all users who upgraded to udev/0.124-1 or later.

I unfortunately had pinned udev to version 0.114-2 while waiting that
bug #493075 is fixed in testing: that's why I encountered the issue
with chrony and the new kernel...

Maybe chrony could recommend udev (>= 0.124-1) ?

--

-- 
 http://frx.netsons.org/doc/index.html#nanodocs
 The nano-document series is here!
..................................................... Francesco Poli .
 GnuPG key fpr == C979 F34B 27CE 5CD8 DC12  31B5 78F4 279B DD6D FCF4
Permalink | Reply |

Gmane