headers
Alex Goncharov | 1 Sep 2008 01:32
Picon

Re: WARNING: 7-STABLE BROKEN -- please wait to upgrade

,--- You/John (Sun, 31 Aug 2008 22:36:36 +0000) ----*
| If you still have the obj tree from the problem build, there are two things
| you can do:
| 
| 1. make STRIP= installworld

Thank you -- that's good to know, for the future.

For now, I updated again and using my saved "four cc tools kit"
rebuilt the world and kernel.  After another reboot everything was
fine -- I can use the new `cc' all right now.

| or
| 
| 2. Delete the obj tree before building anything new. 
| 
| > Is everything supposed to work out of box now?
| 
| Yes, but an obj tree from a broken build will cause problems.

That's a bit strange:

1. What in it could cause problems?  Doesn't `buildworld' clean out
   everything in the `obj' tree, other than the "build counter"?

2. I did my rebuild (after the message I sent out this morning),
   without touching `obj' by hand -- and everything came out all
   right.

(Just wondering)
(Continue reading)

Permalink | Reply | 2 comments |
headers
John Birrell | 1 Sep 2008 01:35

Re: WARNING: 7-STABLE BROKEN -- please wait to upgrade

On Sun, Aug 31, 2008 at 07:32:14PM -0400, Alex Goncharov wrote:
> | or
> | 
> | 2. Delete the obj tree before building anything new. 
> | 
> | > Is everything supposed to work out of box now?
> | 
> | Yes, but an obj tree from a broken build will cause problems.
> 
> That's a bit strange:
> 
> 1. What in it could cause problems?  Doesn't `buildworld' clean out
>    everything in the `obj' tree, other than the "build counter"?

It's supposed to, but I don't trust it. There is no substitute for
a rm -rf of the obj tree. That's also quicker.

> 2. I did my rebuild (after the message I sent out this morning),
>    without touching `obj' by hand -- and everything came out all
>    right.

If that works for you, great.

Sorry about the problems.

--
Johb Birrell
_______________________________________________
freebsd-stable <at> freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
(Continue reading)

Permalink | Reply | 1 comment |
headers
Alex Goncharov | 1 Sep 2008 01:39
Picon

Re: WARNING: 7-STABLE BROKEN -- please wait to upgrade

,--- You/John (Sun, 31 Aug 2008 23:35:38 +0000) ----*
| It's supposed to, but I don't trust it. There is no substitute for
| a rm -rf of the obj tree. That's also quicker.

Ah, OK...

| Sorry about the problems.

Oh, no problem at all -- I learned a lot as a result.  Just wanted to
make sure I understand the state of affairs and ways to act.

Thank you!

-- Alex -- alex-goncharov <at> comcast.net --
_______________________________________________
freebsd-stable <at> freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "freebsd-stable-unsubscribe <at> freebsd.org"
Permalink | Reply | 0 comments |
headers
John Birrell | 1 Sep 2008 03:52

Re: Unresponsive after dtrace

On Sun, Aug 31, 2008 at 06:35:16AM -0500, Wes Morgan wrote:
> I know this has been reported already, but I want to give a "me too". 
> After installing a new world and kernel from the tree yesterday afternoon, 
> I let my system run all night. This morning everything was extremely 
> sluggish and unresponsive. According to top, which I thankfully left 
> running, processes were going in and out of "*kmem_" (obviously 
> truncated). CPU usage was 80+% system and load averages were around 5.4. 
> The only changes I made to my system besides upgrading were to include the 
> options KDB, DDB and STACK in my kernel for zfs functionality. 
> Unfortunately, I cannot try without those options since my root is zfs. 
> Booting a kernel from 8/20 works fine.

Wes, will you please try removing "#define KMEM_DEBUG" from:

src/sys/cddl/compat/opensolaris/kern/opensolaris_kmem.c

This is the likely cause of the performance problems you are seeing. It
is also the reason why you needed to add DDB, DBB and stack to the kernel.

The quickest way to try this is to just build the 'opensolaris' kernel module:

(keep a copy of your current /boot/kernel)
cd src/sys/modules/opensolaris
make obj && make depend && make all && make install

--
John Birrell
_______________________________________________
freebsd-stable <at> freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
(Continue reading)

Permalink | Reply | 1 comment |
headers
John Birrell | 1 Sep 2008 04:15

ZFS performance issues (solved?!)

For those people experiencing a performance degradation since the DTrace import,
please update your copy of src/sys/cddl/compat/opensolaris/kern/opensolaris_kmem.c
by either cvsup of direct edit to remove "#define KMEM_DEBUG".

You only need to rebuild the opensolaris kernel module after this change. The code
is shared between ZFS and DTrace via the opensolaris kernel module.

This is also the reason why you found it necessary to add KDB, DDB and STACK to
your kernel. After removing KMEM_DEBUG, you won't need those.

Please confirm that this solves the problem you have been seeing.

--
John Birrell
_______________________________________________
freebsd-stable <at> freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "freebsd-stable-unsubscribe <at> freebsd.org"
Permalink | Reply | 5 comments |
headers
Michael | 1 Sep 2008 04:12
Picon

Re: bin/121684: : dump(8) frequently hangs

Any progress here? Does anyone know if this will be fixed in 7.1 latest,
or should we start looking for different backup solution (in this case I
would suggest to remove dump from the source tree - having a backup tool
that doesn't work is worse than having none). After upgrading we
basically cannot backup our servers.

Shouldn't this issue be on http://www.freebsd.org/releases/7.0R/errata.html?

cheers
michael

_______________________________________________
freebsd-stable <at> freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "freebsd-stable-unsubscribe <at> freebsd.org"
Permalink | Reply | 501 comments |
headers
Wes Morgan | 1 Sep 2008 05:45

Re: Unresponsive after dtrace

On Mon, 1 Sep 2008, John Birrell wrote:

> On Sun, Aug 31, 2008 at 06:35:16AM -0500, Wes Morgan wrote:
>> I know this has been reported already, but I want to give a "me too".
>> After installing a new world and kernel from the tree yesterday afternoon,
>> I let my system run all night. This morning everything was extremely
>> sluggish and unresponsive. According to top, which I thankfully left
>> running, processes were going in and out of "*kmem_" (obviously
>> truncated). CPU usage was 80+% system and load averages were around 5.4.
>> The only changes I made to my system besides upgrading were to include the
>> options KDB, DDB and STACK in my kernel for zfs functionality.
>> Unfortunately, I cannot try without those options since my root is zfs.
>> Booting a kernel from 8/20 works fine.
>
> Wes, will you please try removing "#define KMEM_DEBUG" from:
>
> src/sys/cddl/compat/opensolaris/kern/opensolaris_kmem.c
>
> This is the likely cause of the performance problems you are seeing. It
> is also the reason why you needed to add DDB, DBB and stack to the kernel.
>
> The quickest way to try this is to just build the 'opensolaris' kernel module:
>
> (keep a copy of your current /boot/kernel)
> cd src/sys/modules/opensolaris
> make obj && make depend && make all && make install

Everything compiled with my original kern config (no STACK etc). I'll let 
it go for a bit and see, but I don't see any processes going into "kmem_" 
in top yet.
(Continue reading)

Permalink | Reply | 0 comments |
headers
FreeBSD Tinderbox | 1 Sep 2008 08:47
Picon
Favicon

[releng_7 tinderbox] failure on i386/i386

TB --- 2008-09-01 05:28:34 - tinderbox 2.3 running on freebsd-stable.sentex.ca
TB --- 2008-09-01 05:28:34 - starting RELENG_7 tinderbox run for i386/i386
TB --- 2008-09-01 05:28:34 - cleaning the object tree
TB --- 2008-09-01 05:29:00 - cvsupping the source tree
TB --- 2008-09-01 05:29:00 - /usr/bin/csup -r 3 -g -L 1 -h localhost -s /tinderbox/RELENG_7/i386/i386/supfile
TB --- 2008-09-01 05:29:08 - building world (CFLAGS=-O2 -pipe)
TB --- 2008-09-01 05:29:08 - cd /src
TB --- 2008-09-01 05:29:08 - /usr/bin/make -B buildworld
>>> World build started on Mon Sep  1 05:29:10 UTC 2008
>>> Rebuilding the temporary build tree
>>> stage 1.1: legacy release compatibility shims
>>> stage 1.2: bootstrap tools
>>> stage 2.1: cleaning up the object tree
>>> stage 2.2: rebuilding the object tree
>>> stage 2.3: build tools
>>> stage 3: cross tools
>>> stage 4.1: building includes
>>> stage 4.2: building libraries
>>> stage 4.3: make dependencies
>>> stage 4.4: building everything
>>> World build completed on Mon Sep  1 06:35:39 UTC 2008
TB --- 2008-09-01 06:35:39 - generating LINT kernel config
TB --- 2008-09-01 06:35:39 - cd /src/sys/i386/conf
TB --- 2008-09-01 06:35:39 - /usr/bin/make -B LINT
TB --- 2008-09-01 06:35:39 - building LINT kernel (COPTFLAGS=-O2 -pipe)
TB --- 2008-09-01 06:35:39 - cd /src
TB --- 2008-09-01 06:35:39 - /usr/bin/make -B buildkernel KERNCONF=LINT
>>> Kernel build for LINT started on Mon Sep  1 06:35:39 UTC 2008
>>> stage 1: configuring the kernel
>>> stage 2.1: cleaning up the object tree
(Continue reading)

Permalink | Reply | 0 comments |
headers
Henri Hennebert | 1 Sep 2008 09:38
Picon
Favicon

Re: ZFS performance issues (solved?!)

John Birrell wrote:
> For those people experiencing a performance degradation since the DTrace import,
> please update your copy of src/sys/cddl/compat/opensolaris/kern/opensolaris_kmem.c
> by either cvsup of direct edit to remove "#define KMEM_DEBUG".
> 
> You only need to rebuild the opensolaris kernel module after this change. The code
> is shared between ZFS and DTrace via the opensolaris kernel module.
> 
> This is also the reason why you found it necessary to add KDB, DDB and STACK to
> your kernel. After removing KMEM_DEBUG, you won't need those.
> 
> Please confirm that this solves the problem you have been seeing.

Great, now everything is back to normal

Thanks

Henri
> 
> --
> John Birrell
> _______________________________________________
> freebsd-stable <at> freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-stable
> To unsubscribe, send any mail to "freebsd-stable-unsubscribe <at> freebsd.org"

_______________________________________________
freebsd-stable <at> freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "freebsd-stable-unsubscribe <at> freebsd.org"
(Continue reading)

Permalink | Reply | 0 comments |
headers
foreverterran | 1 Sep 2008 09:45
Picon

Linux/Unix system admin looking for job

Hello,
ABOUT ME:

EXPERIENCE SUMMARY:

Operating Systems:
. FreeBSD
. RHEL
. Suse, SLES 9/10
. CentOS
. Fedora
. Ubuntu
. Debian
. Microsoft Windows 98/xp/2000/2003/NT

Software and Networking:
. Apache HTTP/HTTPS servers (versions 1.3.x/2.x )
. Mail (Sendmail, Postfix, Amavisd-new, ClamAV, Spamassassin, Courier IMAP)
. MySQL, PostgreSQL, Sybase (backup, maintenance, tuning, replication)
. Monitoring (MRTG, Zabbix, Munin)
. Samba, NFS
. NTP, ntpdate
. Firewall (iptables, ipfw, ipfilter, shorewall)
. LDAP (OpenLDAP)
. Backup (Rsnapshot, Amanda)
. Virtualisation (Qemu, Vmware, Xen)
. High Availability (Hearbeat, load balancing)
. TCP/IP, UDP, ICMP, DHCP, SNMP, NAT, VPN
. Perl, bash scripting
(Continue reading)

Permalink | Reply | 0 comments |

Gmane