Re: [linux-lvm] problem with lvcreate and redirection of stdout/stderr

Re: [linux-lvm] problem with lvcreate and redirection of stdout/stderr

[Lentes, Bernd]

Alasdair wrote:

On Thu, Aug 21, 2014 at 11:57:51AM -0500, Roger Heflin wrote:
> I can see how with the vg suspended that might have a problem.

> But nothing should be getting written to stderr while the device is suspended.


Hi,

i don't understand what you mean with "should be getting written". Does that mean "it's not possible that something is written" or "it would be better if nothing is written" ?

Bernd

_______________________________________________
linux-lvm mailing list
linux-lvm@redhat.com
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/

Helmholtz Zentrum München
Deutsches Forschungszentrum für Gesundheit und Umwelt (GmbH)
Ingolstädter Landstr. 1
85764 Neuherberg
www.helmholtz-muenchen.de
Aufsichtsratsvorsitzende: MinDir´in Bärbel Brumme-Bothe
Geschäftsführer: Prof. Dr. Günther Wess, Dr. Nikolaus Blum, Dr. Alfons Enhsen
Registergericht: Amtsgericht München HRB 6466
USt-IdNr: DE 129521671

Re: [linux-lvm] problem with lvcreate and redirection of stdout/stderr

[Alasdair G Kergon]

On Fri, Aug 22, 2014 at 11:02:32AM +0200, Lentes, Bernd wrote:
> On Thu, Aug 21, 2014 at 11:57:51AM -0500, Roger Heflin wrote:
> > I can see how with the vg suspended that might have a problem.
> > But nothing should be getting written to stderr while the device is suspended.
 
> i don't understand what you mean with "should be getting written". Does that mean "it's not possible that something is written" or "it would be better if nothing is written" ?
 
The lvm2 code does not write to stderr in that part of the code.
(If that was happening somehow it would be a bug.)

Alasdair

[linux-lvm] problem with lvcreate and redirection of stdout/stderr

[Lentes, Bernd]

Hi ML,

i'm writing a script to update my servers. Before updating, i'd like to create a snapshot of the root-directory (/). If something goes wrong with the update, I have a consistent backup.
I try to redirect stdout and stderr to a logfile for being able to check everything afterwards. The redirection has problems with lvcreate.
This is my script (excerpt):

============================================================
lvremove -fv /dev/vg1/lv_root_snapshot &> /var/log/update.log

lvcreate -v -L 45G -n lv_root_snapshot -s vg1/lv_root >> /var/log/update.log 2>&1

zypper -n up -y -t patch  --skip-interactive >> /var/log/update.log 2>&1
============================================================

With the second line I try to create the lv and redirect stdout/stderr to a log file.
The whole system stop by executing the second line, this are the last lines in my log:

============================================================
lvcreate    Setting logging type to disk
lvcreate    Setting chunksize to 8 sectors.
lvcreate    Finding volume group "vg1"
lvcreate    Archiving volume group "vg1" metadata (seqno 99).
lvcreate    Creating logical volume lv_root_snapshot
lvcreate    Creating volume group backup "/etc/lvm/backup/vg1" (seqno 100).
lvcreate    Found volume group "vg1"
lvcreate    activation/volume_list configuration setting not defined: Checking only host tags for vg1/lv_root_snapshot
lvcreate    Creating vg1-lv_root_snapshot
lvcreate    Loading vg1-lv_root_snapshot table (252:1)
lvcreate    Resuming vg1-lv_root_snapshot (252:1)
lvcreate    Clearing start of logical volume "lv_root_snapshot"
lvcreate    Creating logical volume snapshot0
lvcreate    Found volume group "vg1"
lvcreate    Found volume group "vg1"
lvcreate    Creating vg1-lv_root-real
lvcreate    Loading vg1-lv_root-real table (252:2)
lvcreate    Loading vg1-lv_root table (252:0)
lvcreate    Creating vg1-lv_root_snapshot-cow
lvcreate    Loading vg1-lv_root_snapshot-cow table (252:3)
lvcreate    Resuming vg1-lv_root_snapshot-cow (252:3)
lvcreate    Loading vg1-lv_root_snapshot table (252:1)
lvcreate    Suspending vg1-lv_root (252:0) with filesystem sync with device flush
===========================================================

I tried it several times, and the system always stop at this point. When I execute the same command manually in a shell no change. Stop.
Redirecting the lvremove command on a shell in the same manner is working fine:

lvremove -fv /dev/vg1/lv_root_snapshot >> /var/log/update.log 2>&1
system continues running.

lvcreate manually in a shell and only redirecting stdout is working fine:

lvcreate -v -L 45G -n lv_root_snapshot -s vg1/lv_root >> /var/log/update.log
system continues running

Only the combination of redirecting stdout and stderr to the logfile stops the system completely:
lvcreate -v -L 45G -n lv_root_snapshot -s vg1/lv_root >> /var/log/update.log 2>&1


Any ideas ?


my system:

SLES 11 SP3 64bit
vm58820-6:~ # lvm version
  LVM version:     2.02.98(2) (2012-10-15)
  Library version: 1.03.01 (2011-10-15)
  Driver version:  4.23.0
1 lv for the root-directory, /boot is a "real" partition
1pv, 1 vg

Bernd


--
Bernd Lentes

Systemadministration
Institut f�r Entwicklungsgenetik
Geb�ude 35.34 - Raum 208
HelmholtzZentrum m�nchen
bernd.lentes@helmholtz-muenchen.de
phone: +49 89 3187 1241
fax:   +49 89 3187 2294
http://www.helmholtz-muenchen.de/idg

Die Freiheit wird nicht durch weniger Freiheit verteidigt



Helmholtz Zentrum M�nchen
Deutsches Forschungszentrum f�r Gesundheit und Umwelt (GmbH)
Ingolst�dter Landstr. 1
85764 Neuherberg
www.helmholtz-muenchen.de
Aufsichtsratsvorsitzende: MinDir�in B�rbel Brumme-Bothe
Gesch�ftsf�hrer: Prof. Dr. G�nther Wess, Dr. Nikolaus Blum, Dr. Alfons Enhsen
Registergericht: Amtsgericht M�nchen HRB 6466
USt-IdNr: DE 129521671

Re: [linux-lvm] problem with lvcreate and redirection of stdout/stderr

[Alasdair G Kergon]

On Wed, Aug 20, 2014 at 05:27:52PM +0200, Lentes, Bernd wrote:
> lvcreate -v -L 45G -n lv_root_snapshot -s vg1/lv_root >> /var/log/update.log 2>&1
 
> lvcreate    Suspending vg1-lv_root (252:0) with filesystem sync with device flush

/var/log is presumably part of a filesystem on the same LV that is getting its I/O
paused for a brief moment while the snapshot is taken.

1) Run lvm dumpconfig log/activation  to double-check this is not set to 1.

2) What filesystem is this and precisely what kernel?  (cat /proc/mounts)

3) Try to obtain a stack trace at the point where this locks up e.g.  
echo t > /proc/sysrq-trigger followed by dmesg (or divert your logging
of kernel messages to a different filesystem outside the snapshot).

Alasdair

Re: [linux-lvm] problem with lvcreate and redirection of stdout/stderr

[Lentes, Bernd]

Alasdair wrote:

>
> On Wed, Aug 20, 2014 at 05:27:52PM +0200, Lentes, Bernd wrote:
> > lvcreate -v -L 45G -n lv_root_snapshot -s vg1/lv_root >>
> > /var/log/update.log 2>&1
>
> > lvcreate    Suspending vg1-lv_root (252:0) with filesystem sync with device
> flush
>
> /var/log is presumably part of a filesystem on the same LV that is getting its
> I/O paused for a brief moment while the snapshot is taken.
>

Yes it is. I have just one lv for the whole directory / .

> 1) Run lvm dumpconfig log/activation  to double-check this is not set to 1.

log {
        verbose=2
        syslog=0
        file="/var/log/lvm2.log"
        overwrite=0
        level=5
        indent=1
        command_names=1
        prefix="  "
}

activation {
        missing_stripe_filler="/dev/ioerror"
        reserved_stack=256
        reserved_memory=8192
        process_priority=-18
        mirror_region_size=512
        readahead="auto"
        monitoring=1
        mirror_log_fault_policy="allocate"
        mirror_device_fault_policy="remove"
        thin_pool_autoextend_threshold=100
        thin_pool_autoextend_percent=20
        udev_rules=1
        udev_sync=1
}


>
> 2) What filesystem is this and precisely what kernel?  (cat /proc/mounts)

FS is ext3, kernel is 3.0.76-0.11-default

vm58820-6:~ # cat /proc/mounts
rootfs / rootfs rw 0 0
udev /dev tmpfs rw,relatime,nr_inodes=0,mode=755 0 0
tmpfs /dev/shm tmpfs rw,relatime 0 0
/dev/mapper/vg1-lv_root / ext3 rw,relatime,errors=continue,user_xattr,acl,barrier=1,data=ordered 0 0
proc /proc proc rw,relatime 0 0
sysfs /sys sysfs rw,relatime 0 0
devpts /dev/pts devpts rw,relatime,gid=5,mode=620,ptmxmode=000 0 0
debugfs /sys/kernel/debug debugfs rw,relatime 0 0
/dev/vda3 /boot ext3 rw,relatime,errors=continue,user_xattr,acl,barrier=1,data=ordered 0 0
fusectl /sys/fs/fuse/connections fusectl rw,relatime 0 0
securityfs /sys/kernel/security securityfs rw,relatime 0 0
none /var/lib/ntp/proc proc ro,nosuid,nodev,relatime 0 0


>
> 3) Try to obtain a stack trace at the point where this locks up e.g.
> echo t > /proc/sysrq-trigger followed by dmesg (or divert your logging of
> kernel messages to a different filesystem outside the snapshot).
>
> Alasdair

the stack trace you find here (~190KB):  http://www.filedropper.com/stacktracevm58820-6lvcreate

Strange thing is that it worked now once without system stop. But second time stop again.


Thanks for your help.



Bernd

Helmholtz Zentrum München
Deutsches Forschungszentrum für Gesundheit und Umwelt (GmbH)
Ingolstädter Landstr. 1
85764 Neuherberg
www.helmholtz-muenchen.de
Aufsichtsratsvorsitzende: MinDir´in Bärbel Brumme-Bothe
Geschäftsführer: Prof. Dr. Günther Wess, Dr. Nikolaus Blum, Dr. Alfons Enhsen
Registergericht: Amtsgericht München HRB 6466
USt-IdNr: DE 129521671

Re: [linux-lvm] problem with lvcreate and redirection of stdout/stderr

[Roger Heflin]

stderr flushes on each write of an line to attempt to not lose data.

I can see how with the vg suspended that might have a problem.
Everything will stop waiting for the stderr flush to complete, ie the
print statement doing the output to standard error will never return
and then execute the next command to continue on with the vg work.

I would suggest sending the output data to another device (maybe a
tmpfs device) then when done copy it back.

On Wed, Aug 20, 2014 at 12:20 PM, Lentes, Bernd
<bernd.lentes@helmholtz-muenchen.de> wrote:
> Alasdair wrote:
>
>>
>> On Wed, Aug 20, 2014 at 05:27:52PM +0200, Lentes, Bernd wrote:
>> > lvcreate -v -L 45G -n lv_root_snapshot -s vg1/lv_root >>
>> > /var/log/update.log 2>&1
>>
>> > lvcreate    Suspending vg1-lv_root (252:0) with filesystem sync with device
>> flush
>>
>> /var/log is presumably part of a filesystem on the same LV that is getting its
>> I/O paused for a brief moment while the snapshot is taken.
>>
>
> Yes it is. I have just one lv for the whole directory / .
>
>> 1) Run lvm dumpconfig log/activation  to double-check this is not set to 1.
>
> log {
>         verbose=2
>         syslog=0
>         file="/var/log/lvm2.log"
>         overwrite=0
>         level=5
>         indent=1
>         command_names=1
>         prefix="  "
> }
>
> activation {
>         missing_stripe_filler="/dev/ioerror"
>         reserved_stack=256
>         reserved_memory=8192
>         process_priority=-18
>         mirror_region_size=512
>         readahead="auto"
>         monitoring=1
>         mirror_log_fault_policy="allocate"
>         mirror_device_fault_policy="remove"
>         thin_pool_autoextend_threshold=100
>         thin_pool_autoextend_percent=20
>         udev_rules=1
>         udev_sync=1
> }
>
>
>>
>> 2) What filesystem is this and precisely what kernel?  (cat /proc/mounts)
>
> FS is ext3, kernel is 3.0.76-0.11-default
>
> vm58820-6:~ # cat /proc/mounts
> rootfs / rootfs rw 0 0
> udev /dev tmpfs rw,relatime,nr_inodes=0,mode=755 0 0
> tmpfs /dev/shm tmpfs rw,relatime 0 0
> /dev/mapper/vg1-lv_root / ext3 rw,relatime,errors=continue,user_xattr,acl,barrier=1,data=ordered 0 0
> proc /proc proc rw,relatime 0 0
> sysfs /sys sysfs rw,relatime 0 0
> devpts /dev/pts devpts rw,relatime,gid=5,mode=620,ptmxmode=000 0 0
> debugfs /sys/kernel/debug debugfs rw,relatime 0 0
> /dev/vda3 /boot ext3 rw,relatime,errors=continue,user_xattr,acl,barrier=1,data=ordered 0 0
> fusectl /sys/fs/fuse/connections fusectl rw,relatime 0 0
> securityfs /sys/kernel/security securityfs rw,relatime 0 0
> none /var/lib/ntp/proc proc ro,nosuid,nodev,relatime 0 0
>
>
>>
>> 3) Try to obtain a stack trace at the point where this locks up e.g.
>> echo t > /proc/sysrq-trigger followed by dmesg (or divert your logging of
>> kernel messages to a different filesystem outside the snapshot).
>>
>> Alasdair
>
> the stack trace you find here (~190KB):  http://www.filedropper.com/stacktracevm58820-6lvcreate
>
> Strange thing is that it worked now once without system stop. But second time stop again.
>
>
> Thanks for your help.
>
>
>
> Bernd
>
> Helmholtz Zentrum M�nchen
> Deutsches Forschungszentrum f�r Gesundheit und Umwelt (GmbH)
> Ingolst�dter Landstr. 1
> 85764 Neuherberg
> www.helmholtz-muenchen.de
> Aufsichtsratsvorsitzende: MinDir�in B�rbel Brumme-Bothe
> Gesch�ftsf�hrer: Prof. Dr. G�nther Wess, Dr. Nikolaus Blum, Dr. Alfons Enhsen
> Registergericht: Amtsgericht M�nchen HRB 6466
> USt-IdNr: DE 129521671
>
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@redhat.com
> https://www.redhat.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/

Re: [linux-lvm] problem with lvcreate and redirection of stdout/stderr

[Alasdair G Kergon]

On Thu, Aug 21, 2014 at 11:57:51AM -0500, Roger Heflin wrote:
> I can see how with the vg suspended that might have a problem.
 
But nothing should be getting written to stderr while the device is suspended.

Alasdair