openrc support thread

Message

darkphader · Post by **darkphader** » Sun Apr 27, 2008 3:32 pm

revilootneg wrote:

If you wish to check the consistency of an XFS filesystem or
repair a damaged filesystem, see xfs_check(8) and xfs_repair(8).
If you wish to check the consistency of an XFS filesystem or
repair a damaged filesystem, see xfs_check(8) and xfs_repair(8).
If you wish to check the consistency of an XFS filesystem or
repair a damaged filesystem, see xfs_check(8) and xfs_repair(8).

I was getting that also but decided to remerge baselayout and openrc this morning after seeing your post and now I no longer see that issue. Looks like some fixes were made without bumping the release version.

gerard27 · Post by **gerard27** » Sun Apr 27, 2008 4:39 pm

Hi,
Well I guess I'll have to live with irqbalance not deamonizing.
alex.blackbit's question hasn't been answered so far.I'd like to know also.

There is one more thing I notice:
During boot I get this message just prior to baselayout starting:

Code: Select all

* Waitpid:Interrupted system call.

The asterisk is red.
I searched the forums but only found an Italian item with the same error.
I can't read Italian,but it had to do with the new baselayout.
A search on the net only came up with posts dating back to 2004.
I don't get this message on the old baselayout
And /proc/interrupts looks better on the old one.
Gerard.

Cyker · Post by **Cyker** » Sun Apr 27, 2008 4:49 pm

I wouldn't worry about irqbalance that much.

On single-socket systems, it makes /proc/interrupts look nicer; That's all.

The IRQ lines still goto that socket and to the two cores via a common interface, so directing them to one or the other won't make a lick of difference.

gerard27 · Post by **gerard27** » Sun Apr 27, 2008 6:11 pm

Thanks Cyker.
Gerard.

UberLord · Post by **UberLord** » Sun Apr 27, 2008 9:37 pm

dR0PS wrote:I wonder if devs could tell me an appropriate place for placing tunctl ?
When using an inkernel tun/tap and not a kernel module, the final device must be created through tunctl.

You should read net.example found in /usr/share/doc/openrc - it has a section for configuring tun/tap devices.

UberLord · Post by **UberLord** » Sun Apr 27, 2008 9:37 pm

alex.blackbit wrote:UberLord, could you please describe what difference it makes to baselayout whether i have to cores in one socket or two seperated cpu's?

baselayout / openrc doesn't care about such things.

UberLord · Post by **UberLord** » Sun Apr 27, 2008 9:43 pm

Gerard van Vuuren wrote:Well I guess I'll have to live with irqbalance not deamonizing.
alex.blackbit's question hasn't been answered so far.I'd like to know also.

You should file a bug with the irqbalance team to add a switch to test if irqbalance will daemonise. Then we get best of both worlds. For example

if irqbalance --will-daemonise; then
start-stop-daemon --start --exec /path/to/irqbalance
else
irqbalance
fi

If not, then just don't use start-stop-daemon. But then you loose the nice daemon tracking it provides.

There is one more thing I notice:
During boot I get this message just prior to baselayout starting:
Code: Select all
* Waitpid:Interrupted system call.
The asterisk is red.

http://bugs.gentoo.org/show_bug.cgi?id=219179

And I have no idea why this happens, nor can I replicate it :/

alex.blackbit · Post by **alex.blackbit** » Mon Apr 28, 2008 2:11 pm

UberLord wrote:irqbalance only works on SMP systems.
Multicore is NOT SMP according to irqbalance.

So, how many CPU's do you have?

UberLord wrote:
alex.blackbit wrote:UberLord, could you please describe what difference it makes to baselayout whether i have to cores in one socket or two seperated cpu's?
baselayout / openrc doesn't care about such things.

so, that confuses me.
or is irqbalance a completely different story? how should it even know if the cpu cores available to the system are from a multi-core or multi-socket system?

UberLord · Post by **UberLord** » Mon Apr 28, 2008 2:20 pm

ok here we go

OpenRC / baselayout don't give two hoots how many CPU's, cores or HT you have. The code is the same.
However, pertinent to your query, start-stop-daemon cares that daemons it starts are still running for a short time after daemonising.

What's happening is that irqbalance is not daemonising and OpenRC is now seeing this and flagging an error.

f0rk · Post by **f0rk** » Mon Apr 28, 2008 2:42 pm

I also have "* Waitpid:Interrupted system call" error in 0.2.2 version.
openrc-9999 doesn't work at all. Read-only error as here: https://bugs.gentoo.org/show_bug.cgi?id=219179

UberLord · Post by **UberLord** » Mon Apr 28, 2008 2:51 pm

And until I get more information there's nothing I can do to fix it :/

gerard27 · Post by **gerard27** » Mon Apr 28, 2008 3:33 pm

Hi,
I wrote down some of the text at the point where the waitpid message appears.

Code: Select all

kjournald starting.Commit interval 5 seconds.
EXT3-fs:Mounted filesystem with ordered data mode.
VFS:Mounted root (ext3 filesystem) readonly.
Freeing ubused kernel memory:224k freed.
INIT:version 2.86 booting.
*Waitpid:Interrupted system call
OpenRC 0.2.2 is starting up Gentoo Linux (i686)
Press I to enter interactive mode.

I browsed sbin/init and it has "waitpid" in it.
I also browsed the C sources of openrc and it has it in several places.
Please remember I have 2 root partitions with identical binaries.
The only difference is baselayout.
With the old baselayout I have no problems.
Gerard.

UberLord · Post by **UberLord** » Mon Apr 28, 2008 4:10 pm

For those with the waitpid error, could you try the git ebuild please?
rc --version should report git-857b2c9f

f0rk · Post by **f0rk** » Mon Apr 28, 2008 5:21 pm

UberLord wrote:For those with the waitpid error, could you try the git ebuild please?
rc --version should report git-857b2c9f

Good job, it works well! Thanks a lot!

gerard27 · Post by **gerard27** » Mon Apr 28, 2008 6:01 pm

Hi UberLord,
That's what the doctor ordered.No more waitpid error.
When will it go stable?
Gerard.

darkphader · Post by **darkphader** » Mon Apr 28, 2008 7:09 pm

darkphader wrote:...issues with openrc.
alsa settings don't save and restore
nsd doesn't start, which means that:
named doesn't start
mit-krb5kdc doesn't start
mit-krb5kadmind doesn't start
txquelen doesn't get set (oddly enough mtu does)

Decided to go with the git version and with it the nsd/named/mit* problems above are resolved.
The alsa issue was not a baselayout/openrc problem, the mixer was getting saved and restored, logging into kde was changing it.

The setting of txqueuelen is still borked.
https://bugs.gentoo.org/show_bug.cgi?id=217948
To add: this bug exists on both x86 and x86_64.

revilootneg · Post by **revilootneg** » Tue Apr 29, 2008 7:58 am

darkphader wrote: I was getting that also but decided to remerge baselayout and openrc this morning after seeing your post and now I no longer see that issue. Looks like some fixes were made without bumping the release version.

darkphader:
Unfortunately your tip didn't work for me (using openrc-0.2.2 / git master with 0.2.3). Did you refer to openrc-0.2.2 or openrc from git?

Uberlord:
Any tips regarding
- the xfs-fsck error messages
- that lvm-after-fsck issue EDIT: fixed in git
- lvm locking error message? EDIT2: done for me too, see below!
(See my post on page seven for details)

EDIT: Tried git-version and can say that the lvm-after-fsck issue is fixed.
EDIT2: Found solution here, point 5. I've /var on a separate partition and of course it's not mounted.

Thanks
revilootneg

UberLord · Post by **UberLord** » Tue Apr 29, 2008 8:50 am

Regarding the xfs error.
http://bugs.gentoo.org/show_bug.cgi?id=209296

revilootneg · Post by **revilootneg** » Tue Apr 29, 2008 9:58 am

UberLord,
will upgrade to xfsprogs-2.9.8 then.

@lvm:
I was reading through /lib/rcscripts/addons/lvm-start.sh.
Any lvm-related command uses --ignorelockingfailure except for /sbin/pvscan in line 28:

Code: Select all

    #still echo stderr for debugging
    # Extra PV find pass because some devices might not have been available until very recently
    /sbin/pvscan >/dev/null

But this is a lvm2 issue then.
Unfortunately I've no access to the machine in question right now to check it.

revilootneg

darkphader · Post by **darkphader** » Tue Apr 29, 2008 1:38 pm

revilootneg wrote:darkphader:
Unfortunately your tip didn't work for me (using openrc-0.2.2 / git master with 0.2.3). Did you refer to openrc-0.2.2 or openrc from git?

I was referring to 0.2.2, hadn't yet decided to jump to the git version. Also I had been at xfsprogs-2.9.8 since the 23rd. Hmm...

seqizz · Post by **seqizz** » Tue Apr 29, 2008 4:29 pm

keba wrote: I even had to start NetworkManager and dhcdbd manually because they crashed!
Doesn't anyone have a solution or at least an idea on what to look for? thx

Same! %100..

Can't connect a network via nm-applet (uses NetworkManager i guess) error is : "can't connect supplicant"

Also boot messages says "Can't started dbus"

UberLord · Post by **UberLord** » Tue Apr 29, 2008 4:36 pm

openrc-0.2.3 is already fixed for reporting dhcdbd status.
i've not tested NetworkManager, and probably won't

dbus starts and works just fine.

openrc-0.2.3 isn't in the tree yet (dunno why), so try the git ebuild to get snapshot.

revilootneg · Post by **revilootneg** » Tue Apr 29, 2008 7:07 pm

Thanks UberLord and darkphader,
xfsprogs-2.9.8 works and adding --ignorelockingfailure to /sbin/pvscan in /lib/rcscripts/addons/lvm-start.sh (line 28) fixes the "Locking error"-message.

Thanks UberLord for openrc. I'm sure quite soon the bios boot-phase will take biggest slice of the boot-time-cake!

Atzeni · Post by **Atzeni** » Tue Apr 29, 2008 10:07 pm

Hi folks,

just smoothly migrated to OpenRC.
One issue only (I am not sure whether it's openrc-related or not):
I get a "file not found" from the switch runlevel script reported in the Power Mgmt Guide (http://www.gentoo.org/doc/en/power-management-guide.xml).

Code: Select all

if on_ac_power
then
    if [[ "$(</var/lib/init.d/softlevel)" != "${RUNLEVEL_AC}" ]]
    then
        logger "Switching to ${RUNLEVEL_AC} runlevel"
         /sbin/rc ${RUNLEVEL_AC}
    fi
elif [[ "$(</var/lib/init.d/softlevel)" != "${RUNLEVEL_BATTERY}" ]]
then
    logger "Switching to ${RUNLEVEL_BATTERY} runlevel"
    /sbin/rc ${RUNLEVEL_BATTERY}
fi

Basically the two if statements fail because there is no /var/lib/init.d/softlevel. Is it BL2/openrc that changed something? Never had this error before the migration.

Thanks

violagirl · Post by **violagirl** » Wed Apr 30, 2008 12:48 am

I have been wondering, the migration guide says..

/etc/conf.d/rc has been deprecated and any settings you have in there will need to be migrated to the appropriate settings in /etc/rc.conf. Please read through /etc/rc.conf and /etc/conf.d/rc and migrate the settings. Once you are complete, delete /etc/conf.d/rc.

Okay.... but what about the random settings not in the new sample rc.conf file that are in the rc file by default? I am not sure what to do with them. For example, some things in my rc I couldn't find an example for in the new rc.conf....

Code: Select all

RC_NET_STRICT_CHECKING="no"
RC_DOWN_INTERFACE="yes"
RC_VOLUME_ORDER="raid evms lvm dm"
RC_VERBOSE="no"
RC_BOOTLOG="no"
RC_BOOTCHART="no"
RC_USE_FSTAB="no"
RC_USE_CONFIG_PROFILE="yes"
RC_FORCE_AUTO="no"
RC_DEVICES="auto"
RC_SWAP_ERASE="no"
RC_RETRY_KILL="yes"
RC_RETRY_TIMEOUT=1
RC_RETRY_COUNT=5
RC_FAIL_ON_ZOMBIE="no"
RC_KILL_CHILDREN="no"
RC_WAIT_ON_START="0.1"
svcdir="/var/lib/init.d"
svcmount="no"
svcfstype="tmpfs"
svcsize=2048

I know some of it, like the stuff about raid, no longer applies, but how can I tell how much of this is supposed to go over? I didn't see examples for anything like this in the rc.conf file, so I wasn't sure if I should assume it's no longer needed... or just move all of it over....

openrc support thread

Re: openrc with lvm and xfs partitions

Re: appropriate place for tunctl in openrc environment ?

Re: openrc with lvm and xfs partitions

Re: openrc with lvm and xfs partitions

Re: netmount problem