Mailing List Archive: Problems with latest CVS vesrion

Problems with latest CVS vesrion

Nov 30, 2000, 7:37 AM

Post #1 of 11 (3084 views)

Hello !
Today I have downloaded and installed the latest CVS version of DRBD.
(and also latest beta heartbeat
I can't say it was very easy, but after 3 hours I'm able to mount
/export from /dev/drbd0 ;))
end test hearbeat.
Well, during installation I have descovered some number of problems and
now list it below:
1. Major number problem
In drbd.c major number defined as 43 (240 commented out) but
Makefile::dev creates drbd0 files
with major 240 I'm not quite sure what I did it in a right way but I
change Makefile to use 43 major
Does it posses any potential threat to the system ?
2. _Probably_ wrong order of arguments for datadisk script.
Both datadisk and drbd_common expect action as first argument and
resource as the second.
If you place datadisk to haresources file like this datadisk::drbd0 it
will be incorrectly called with
arguments order like this datadisk drbd0 start
Solution, not to use drbd0 at all and call just datadisk start, or
create simple script to reversing
argument order and place it to haresources file (as I did)
3. Not actually a bug and could be forgiven in respect for beta state,
but anyway, please add this
to release note - configuration file syntax changed, name of device file
changed (reflect it in your /etc/fstab)
4. what is difference bitween drbd_commOn and drbd_commUn ? drbd script
looks for it in /etc/ha.d/resource.d
but make install doesn't copy it from source directory
5. SAFE variable problem. Where should I place this variable ? I have
placed it to drbd_common, but probably
you have a different opinion on it.
SAFE=true problem. I'm running RedHat 6.2 with bash 1.14.7(1) and read
-n from "safe"
branch in datadisk script produce error msg:
read: unknown option: -n
read: usage: read [-r] [name...]

Just a question...
I'm using reiserfs for drbd device. Should I add "sync" flag to the
fstab to be fail-safe with "B" protocol ?
Thanks a lot !
With best regards.

--
Dim

Re: Problems with latest CVS vesrion [ In reply to ]

thomas.mangin at example

Nov 30, 2000, 10:20 AM

Post #2 of 11 (3044 views)

Permalink

> Today I have downloaded and installed the latest CVS version of DRBD.
> (and also latest beta heartbeat
> I can't say it was very easy, but after 3 hours I'm able to mount
> /export from /dev/drbd0 ;))
> end test hearbeat.

It prove you are perseverant ;*)

> 1. Major number problem
> In drbd.c major number defined as 43 (240 commented out) but
> Makefile::dev creates drbd0 files
> with major 240 I'm not quite sure what I did it in a right way but I
> change Makefile to use 43 major

No my mistake ;*)

> Does it posses any potential threat to the system ?

Don't think so .. You must check a driver is at this location.

> 2. _Probably_ wrong order of arguments for datadisk script.
> Both datadisk and drbd_common expect action as first argument and
> resource as the second.
> If you place datadisk to haresources file like this datadisk::drbd0 it
> will be incorrectly called with
> arguments order like this datadisk drbd0 start
> Solution, not to use drbd0 at all and call just datadisk start, or
> create simple script to reversing
> argument order and place it to haresources file (as I did)

Ok, can you please send me your code for me to fix datadisk.

> 3. Not actually a bug and could be forgiven in respect for beta state,
> but anyway, please add this
> to release note - configuration file syntax changed, name of device file
> changed (reflect it in your /etc/fstab)

I don't understand can you excuse my french ... nationality and be mor
everbose ;*)

> 4. what is difference bitween drbd_commOn and drbd_commUn ? drbd script
> looks for it in /etc/ha.d/resource.d
> but make install doesn't copy it from source directory

Un is to be deleted .. my mistake .. I am not a cvs expert yet ..
I done a "cvs delete drbd_commun" seems it didn't work like I expected it
..

> 5. SAFE variable problem. Where should I place this variable ? I have
> placed it to drbd_common, but probably
> you have a different opinion on it.
> SAFE=true problem. I'm running RedHat 6.2 with bash 1.14.7(1) and read
> -n from "safe"

It is the last code I added perhaps still buggy ;*)
It must be set in the drdb(Node) configuration file
See drbd.example ..

> branch in datadisk script produce error msg:
> read: unknown option: -n
> read: usage: read [-r] [name...]

Thanks ..
It must be bash specific .. What is your linux distribution ?
It work fine under my Mandrake 7.2

If you set GLOBAL_SAFE=false in your config file the problem must diseapear

> Just a question...
> I'm using reiserfs for drbd device. Should I add "sync" flag to the
> fstab to be fail-safe with "B" protocol ?

I don't think reiserfs have the sync option.

Thomas

Re: Problems with latest CVS vesrion [ In reply to ]

zager at example

Nov 30, 2000, 12:04 PM

Post #3 of 11 (3033 views)

Permalink

Thomas Mangin wrote:

> > Today I have downloaded and installed the latest CVS version of DRBD.
> > (and also latest beta heartbeat
> > I can't say it was very easy, but after 3 hours I'm able to mount
> > /export from /dev/drbd0 ;))
> > end test hearbeat.
>
> It prove you are perseverant ;*)
>
> > 1. Major number problem
> > In drbd.c major number defined as 43 (240 commented out) but
> > Makefile::dev creates drbd0 files
> > with major 240 I'm not quite sure what I did it in a right way but I
> > change Makefile to use 43 major
>
> No my mistake ;*)
>
> > Does it posses any potential threat to the system ?
>
> Don't think so .. You must check a driver is at this location.

BTW, I have encountered another problem. Before switching to DRBD I have
played with
kernel's NBD, and forget to remove it's code from kernel after installing
DRBD, and at
startup drbd script complain about it. So, probably major 240 isn't so bad,
since it allows
to have both kernel NBD (why not ?) and DRBD at the same time

> > 2. _Probably_ wrong order of arguments for datadisk script.
> > Both datadisk and drbd_common expect action as first argument and
> > resource as the second.
> > If you place datadisk to haresources file like this datadisk::drbd0 it
> > will be incorrectly called with
> > arguments order like this datadisk drbd0 start
> > Solution, not to use drbd0 at all and call just datadisk start, or
> > create simple script to reversing
> > argument order and place it to haresources file (as I did)
>
> Ok, can you please send me your code for me to fix datadisk.

Actually, this is not a patch to diskdata, just a small fake script
"diskdata_" with one line:
diskdata $2 $1, and I call it from haresources file.
I think you should check drbd_common also, because it performs converting of
arguments to script's internal
variables, and it should be fixed too.

> > 3. Not actually a bug and could be forgiven in respect for beta state,
> > but anyway, please add this
> > to release note - configuration file syntax changed, name of device file
> > changed (reflect it in your /etc/fstab)
>
> I don't understand can you excuse my french ... nationality and be mor
> everbose ;*)

Hehe ... my native language is russian, so excuse me too ;))
Well, more verbose version of above sentence:
During installation of latest DRBD I noticed what configuration file format
was
significantly changed, in other words my old configuration file was no longer
valid.
Also, I have discovered what you have changed default device file name from
/dev/nbN to /dev/drdbN (probably I'm wrong, but with previous version I used
/dev/nb0)
This also caused me to make additional changes to drdb0 config file and
/etc/fstab
All this changes to be discovered by me took some fraction of my time.
Excuse me if I misslooked this facts in ChangeLog and Readme.
I just asking you to reflect this changes in documentation for final 0.5.8
release to help
other people during installation. I'm sure you will add it even without my
remark.

> > 4. what is difference bitween drbd_commOn and drbd_commUn ? drbd script
> > looks for it in /etc/ha.d/resource.d
> > but make install doesn't copy it from source directory
>
> Un is to be deleted .. my mistake .. I am not a cvs expert yet ..
> I done a "cvs delete drbd_commun" seems it didn't work like I expected it
> ..

You also should remove this file from your development directory, I think.
But there is another problem - drdb_common checks for drdb_commun : have_file
$INSTALL/drdb_commun,
and drdb script also does so. Hope this should be very easy fix.

> > 5. SAFE variable problem. Where should I place this variable ? I have
> > placed it to drbd_common, but probably
> > you have a different opinion on it.
> > SAFE=true problem. I'm running RedHat 6.2 with bash 1.14.7(1) and read
> > -n from "safe"
>
> It is the last code I added perhaps still buggy ;*)
> It must be set in the drdb(Node) configuration file
> See drbd.example ..
>
> > branch in datadisk script produce error msg:
> > read: unknown option: -n
> > read: usage: read [-r] [name...]
>
> Thanks ..
> It must be bash specific .. What is your linux distribution ?

RedHat 6.2 with bash 1.14.7(1)

>
> It work fine under my Mandrake 7.2
>
> If you set GLOBAL_SAFE=false in your config file the problem must diseapear
>
> > Just a question...
> > I'm using reiserfs for drbd device. Should I add "sync" flag to the
> > fstab to be fail-safe with "B" protocol ?
>
> I don't think reiserfs have the sync option.

Hm ... does it mean what with reiserfs where is no write cache ?

Re: Problems with latest CVS vesrion [ In reply to ]

dg at example

Nov 30, 2000, 3:09 PM

Post #4 of 11 (3044 views)

Permalink

On Thu, Nov 30, 2000 at 08:04:30PM +0100, Dim Segebart wrote:
> Thomas Mangin wrote:
>
> > > 1. Major number problem
> > > In drbd.c major number defined as 43 (240 commented out) but
> > > Makefile::dev creates drbd0 files
> > > with major 240 I'm not quite sure what I did it in a right way but I
> > > change Makefile to use 43 major
> >
> > No my mistake ;*)
> >
> > > Does it posses any potential threat to the system ?
> >
> > Don't think so .. You must check a driver is at this location.
>
> BTW, I have encountered another problem. Before switching to DRBD I have
> played with
> kernel's NBD, and forget to remove it's code from kernel after installing
> DRBD, and at
> startup drbd script complain about it. So, probably major 240 isn't so bad,
> since it allows
> to have both kernel NBD (why not ?) and DRBD at the same time

No no no. This will not work with current kernels. DRBD uses two request
structures and will deadlock if the kernel does not allow for this. There
is code in ll_rw_block() that handles this, but it only does it for major
number 43. Major 240 is reserved for future use of RBD with 2.4 kernels,
but must not be used with 2.2 kernels.

-dg

--
David Gould dg@example.com
SuSE, Inc., 580 2cd St. #210, Oakland, CA 94607 510.628.3380
"I personally think Unix is "superior" because on LSD it tastes
like Blue." -- jbarnett

Re: Problems with latest CVS vesrion [ In reply to ]

ok at example

Nov 30, 2000, 3:56 PM

Post #5 of 11 (3040 views)

Permalink

On Friday 01 December 2000 04:20, Thomas Mangin wrote:
> > Today I have downloaded and installed the latest CVS version of DRBD.
> > (and also latest beta heartbeat
> > I can't say it was very easy, but after 3 hours I'm able to mount
> > /export from /dev/drbd0 ;))
> > end test hearbeat.
>
> It prove you are perseverant ;*)
>
> > 1. Major number problem
> > In drbd.c major number defined as 43 (240 commented out) but
> > Makefile::dev creates drbd0 files
> > with major 240 I'm not quite sure what I did it in a right way but I
> > change Makefile to use 43 major
>
> No my mistake ;*)
>
> > Does it posses any potential threat to the system ?
>
> Don't think so .. You must check a driver is at this location.
>
> > 2. _Probably_ wrong order of arguments for datadisk script.
> > Both datadisk and drbd_common expect action as first argument and
> > resource as the second.
> > If you place datadisk to haresources file like this datadisk::drbd0 it
> > will be incorrectly called with
> > arguments order like this datadisk drbd0 start
> > Solution, not to use drbd0 at all and call just datadisk start, or
> > create simple script to reversing
> > argument order and place it to haresources file (as I did)
>
> Ok, can you please send me your code for me to fix datadisk.

As far as I can tell, from Rev 1.40 and 1.39, datadisk works like it should,
i.e, datadisk [resource] <action>

You've already made the correct changes Thomas, i.e

RESOURCE=$1
case "$2")

and

drbdc status $RESOURCE

Dim, Can you verify that you have the latest version of datadisk?

> > 3. Not actually a bug and could be forgiven in respect for beta state,
> > but anyway, please add this
> > to release note - configuration file syntax changed, name of device file
> > changed (reflect it in your /etc/fstab)
>
> I don't understand can you excuse my french ... nationality and be mor
> everbose ;*)
>
> > 4. what is difference bitween drbd_commOn and drbd_commUn ? drbd script
> > looks for it in /etc/ha.d/resource.d
> > but make install doesn't copy it from source directory
>
> Un is to be deleted .. my mistake .. I am not a cvs expert yet ..
> I done a "cvs delete drbd_commun" seems it didn't work like I expected it
> ...
>
> > 5. SAFE variable problem. Where should I place this variable ? I have
> > placed it to drbd_common, but probably
> > you have a different opinion on it.
> > SAFE=true problem. I'm running RedHat 6.2 with bash 1.14.7(1) and read
> > -n from "safe"
>
> It is the last code I added perhaps still buggy ;*)
> It must be set in the drdb(Node) configuration file
> See drbd.example ..
>
> > branch in datadisk script produce error msg:
> > read: unknown option: -n
> > read: usage: read [-r] [name...]
>
> Thanks ..
> It must be bash specific .. What is your linux distribution ?
> It work fine under my Mandrake 7.2
>
> If you set GLOBAL_SAFE=false in your config file the problem must diseapear

It doesn't.
There's something weird about that too, I just take out that whole section
otherwise the script works, but gives a little warning.
Otherwise, it works fine under Redhat 6.2 and Mandrake 7.2 for me. (When I
take out the offending section, otherwise, just a harmless warning)

That error is caused by BASH 1.X, either modify it to do something like '-n'
or just take out that section ;-)

Maybe have a if $BASH_VERSION = "1.X"; ...

> > Just a question...
> > I'm using reiserfs for drbd device. Should I add "sync" flag to the
> > fstab to be fail-safe with "B" protocol ?
>
> I don't think reiserfs have the sync option.
>
> Thomas
>
> _______________________________________________
> DRBD-devel mailing list
> DRBD-devel@example.com
> http://lists.sourceforge.net/mailman/listinfo/drbd-devel

--
Omar Kilani
Systems Administrator
Mail Call Couriers Pty Ltd

Re: Problems with latest CVS vesrion [ In reply to ]

zager at example

Dec 1, 2000, 1:18 AM

Post #6 of 11 (3045 views)

Permalink

>
>
> As far as I can tell, from Rev 1.40 and 1.39, datadisk works like it should,
> i.e, datadisk [resource] <action>
>
> You've already made the correct changes Thomas, i.e
>
> RESOURCE=$1
> case "$2")
>
> and
>
> drbdc status $RESOURCE
>
> Dim, Can you verify that you have the latest version of datadisk?
>

Today (1 Dec 2000 ... Hey ! Today is a first winter day :))
export CVSROOT=:pserver:anonymous@example.com:/cvshome/drbd
cvs login
cvs checkout drdb

Content of /drdb/scripts/CVS/Entries

/Makefile/1.11/Mon Nov 27 17:27:45 2000//
/README/1.5/Wed Nov 1 10:29:30 2000//
/TODO/1.3/Mon Nov 27 17:27:45 2000//
/datadisk/1.30/Mon Nov 20 15:51:00 2000//
/datadisk.old/1.2/Wed Sep 20 13:06:28 2000//
/drbd/1.22/Thu Nov 9 09:58:08 2000//
/drbd.example/1.12/Tue Nov 14 19:42:07 2000//
/drbd_common/1.1/Mon Nov 20 15:51:00 2000//
/drbd_commun/1.18/Thu Nov 9 09:58:08 2000//
/drbdc/1.14/Mon Nov 20 15:51:00 2000//

From drdb_common

if [ "$2" != "" ]; then
OPTION=$1
RESOURCE=$2
else
OPTION=$1
RESOURCE=""
fi

From datadisk

# See how we were called.
case "$1" in
start)
if [ -z "$RESOURCE" ]; then
runForAll start
exit $RETVAL
fi

echo SAFE $SAFE

Re: Problems with latest CVS vesrion [ In reply to ]

thomas.mangin at example

Dec 1, 2000, 2:48 AM

Post #7 of 11 (3048 views)

Permalink

>
> As far as I can tell, from Rev 1.40 and 1.39, datadisk works like it
should,
> i.e, datadisk [resource] <action>
>
> You've already made the correct changes Thomas, i.e
>
> RESOURCE=$1
> case "$2")
>
> and
>
> drbdc status $RESOURCE
>
> Dim, Can you verify that you have the latest version of datadisk?

I was doing a parameter test to invert them and forgot to use it ..
Tryed to fix it ..

> > If you set GLOBAL_SAFE=false in your config file the problem must
diseapear
>
> It doesn't.
> There's something weird about that too, I just take out that whole section
> otherwise the script works, but gives a little warning.
> Otherwise, it works fine under Redhat 6.2 and Mandrake 7.2 for me. (When I
> take out the offending section, otherwise, just a harmless warning)
>
> That error is caused by BASH 1.X, either modify it to do something like
'-n'
> or just take out that section ;-)

I will try to fix that too ..
My guess it is a it is bash2 dependant
a #!/bin/bash2 perhasps .. but I don't like that ..

Thomas

Re: Problems with latest CVS vesrion [ In reply to ]

kilani at example

Dec 1, 2000, 2:56 AM

Post #8 of 11 (3043 views)

Permalink

[Warning: This email went through a convoluted time line :-)]
At 09:18 AM 1/12/2000 +0100, Dim Segebart wrote:
> >
> >
> > As far as I can tell, from Rev 1.40 and 1.39, datadisk works like it
> should,
> > i.e, datadisk [resource] <action>
> >
> > You've already made the correct changes Thomas, i.e
> >
> > RESOURCE=$1
> > case "$2")
> >
> > and
> >
> > drbdc status $RESOURCE
> >
> > Dim, Can you verify that you have the latest version of datadisk?
> >
>
>Today (1 Dec 2000 ... Hey ! Today is a first winter day :))
>export CVSROOT=:pserver:anonymous@example.com:/cvshome/drbd
>cvs login
>cvs checkout drdb
>
>Content of /drdb/scripts/CVS/Entries
>
>/Makefile/1.11/Mon Nov 27 17:27:45 2000//
>/README/1.5/Wed Nov 1 10:29:30 2000//
>/TODO/1.3/Mon Nov 27 17:27:45 2000//
>/datadisk/1.30/Mon Nov 20 15:51:00 2000//
>/datadisk.old/1.2/Wed Sep 20 13:06:28 2000//
>/drbd/1.22/Thu Nov 9 09:58:08 2000//
>/drbd.example/1.12/Tue Nov 14 19:42:07 2000//
>/drbd_common/1.1/Mon Nov 20 15:51:00 2000//
>/drbd_commun/1.18/Thu Nov 9 09:58:08 2000//
>/drbdc/1.14/Mon Nov 20 15:51:00 2000//
>
>
> From drdb_common
>
>if [ "$2" != "" ]; then
> OPTION=$1
> RESOURCE=$2
>else
> OPTION=$1
> RESOURCE=""
>fi
>
> From datadisk
>
># See how we were called.
>case "$1" in
> start)
> if [ -z "$RESOURCE" ]; then
> runForAll start
> exit $RETVAL
> fi
>
> echo SAFE $SAFE
>

Should be:

# See how we were called.
RESOURCE=$1
case "$2" in
start)
if [ -z "$RESOURCE" ]; then
runForAll start
exit $RETVAL
fi

And ... skip down to ...

status)
$INSTALL/drbdc $*

and change it to:

$INSTALL/drbdc status $RESOURCE

And all is good (apart from the SAFE thing, which I don't need and just
ripped out. - Looks like you'll need to use bash2)
Although with CVS, ... now I'm getting confused.
Because I'm sure this is in the CVS version, but I may have edited it
manually in that CVS directory.
Maybe I could have CVS access to the scripts directory and commit the
change or something.
Or Thomas can just do it.

/README/1.5/Wed Nov 01 10:29:30 2000//
/datadisk.old/1.2/Wed Sep 20 13:06:28 2000//
/drbd/1.22/Thu Nov 09 09:58:08 2000//
/drbd.example/1.12/Tue Nov 14 19:42:07 2000//
/drbd_commun/1.18/Thu Nov 09 09:58:08 2000//
/datadisk/1.30/Result of merge//
/drbd_common/1.1/Mon Nov 20 15:51:00 2000//
/drbdc/1.14/Tue Nov 21 21:54:59 2000//
/Makefile/1.11/Fri Dec 01 10:11:39 2000//
/TODO/1.3/Fri Dec 01 10:11:39 2000//

Hrm ... OK ... maybe it was my personal changes ? I think CVS merged my own
personal changes into my local tree when I did an 'update'.
(Just confirmed this, and yes, it did.)
Weird.
[.Someone just changed the CVS datadisk to use $ACTION ... try that ... I
can't test it at home ... so I'm not sure wether it actually solves the
problem]

Hope this helps.

Omar

>_______________________________________________
>DRBD-devel mailing list
>DRBD-devel@example.com
>http://lists.sourceforge.net/mailman/listinfo/drbd-devel

Re: Problems with latest CVS vesrion [ In reply to ]

kilani at example

Dec 1, 2000, 3:02 AM

Post #9 of 11 (3039 views)

Permalink

At 09:48 AM 1/12/2000 +0000, Thomas Mangin wrote:
> >
> > As far as I can tell, from Rev 1.40 and 1.39, datadisk works like it
>should,
> > i.e, datadisk [resource] <action>
> >
> > You've already made the correct changes Thomas, i.e
> >
> > RESOURCE=$1
> > case "$2")
> >
> > and
> >
> > drbdc status $RESOURCE
> >
> > Dim, Can you verify that you have the latest version of datadisk?
>
>I was doing a parameter test to invert them and forgot to use it ..
>Tryed to fix it ..

I was wrong.
I made those changes to _my_ drbd CVS tree.
Whoops :-)
Well, it worked for me anyhow :-)

> > > If you set GLOBAL_SAFE=false in your config file the problem must
>diseapear
> >
> > It doesn't.
> > There's something weird about that too, I just take out that whole section
> > otherwise the script works, but gives a little warning.
> > Otherwise, it works fine under Redhat 6.2 and Mandrake 7.2 for me. (When I
> > take out the offending section, otherwise, just a harmless warning)
> >
> > That error is caused by BASH 1.X, either modify it to do something like
>'-n'
> > or just take out that section ;-)
>
>I will try to fix that too ..
>My guess it is a it is bash2 dependant
>a #!/bin/bash2 perhasps .. but I don't like that ..

Well, I don't see how GLOBAL_SAFE is the same as SAFE ?
It's not being declared anywhere, even if you're sourcing the drbdX
configuration files.
Or ... ?

>Thomas
>
>_______________________________________________
>DRBD-devel mailing list
>DRBD-devel@example.com
>http://lists.sourceforge.net/mailman/listinfo/drbd-devel

Re: Problems with latest CVS vesrion [ In reply to ]

kilani at example

Dec 1, 2000, 3:15 AM

Post #10 of 11 (3036 views)

Permalink

Thomas,

I just looked at the modified code, and it just doesn't make sense?
(datadisk)
Usage: $0 [resource] {start|stop|restart|status}

Yet:
(drbd_common)
if [ "$2" != "" ]; then
OPTION=$1
RESOURCE=$2
else
OPTION=$1
RESOURCE=""
fi

I.e OPTION is the resource, and the RESOURCE is the option.

Yet, in datadisk, you go on to test the ACTION:

# See how we were called.
case "$ACTION" in
start)

Which is not defined anywhere ? (Maybe this was suppsoed to be $OPTION)
But then in the $SAFE code (where $SAFE isn't defined anywhere else,
...GLOBAL_SAFE is ...) ACTION is being set to something completely different.

(drbd.example)
# The safe parameter is a security to not allow a system to
# restart after a crash. If set it will log the problem and
# ask for an user action (reboot/continue)

GLOBAL_SAFE=false

...
(datadisk)
ACTION="loop"

while [. $ACTION != "c" -a $ACTION != "C" -a
$ACTION != "r" -a $ACTION != "R" ]; do
read -n 1 -p "Do you want to continue or
reboot [C/R] " ACTION
echo

Now, if say, you adjusted drbd_common as follows:

if [ "$2" != "" ]; then
OPTION=$2
RESOURCE=$1
else
OPTION=$1
RESOURCE=""
fi

It makes a little sense, because then you have RESOURCE defined correctly,
and then in datadisk ...

# See how we were called.
case "$OPTION" in
start)

Then datadisk works correctly.
BUT ...
It makes no sense to allow the user specifying no resource, as in the TODO
you specify that you want datadisk to handle things per drbd device.
Currently, with those changes, it will, and it works very well.

I.e

datadisk start

doesn't make sense.

while datadisk drbd0 start (as it is now ... does.)

Also, should those SAFE references be GLOBAL_SAFE references ?
Or ?

Best Regards,
Omar Kilani

At 09:48 AM 1/12/2000 +0000, Thomas Mangin wrote:
> >
> > As far as I can tell, from Rev 1.40 and 1.39, datadisk works like it
>should,
> > i.e, datadisk [resource] <action>
> >
> > You've already made the correct changes Thomas, i.e
> >
> > RESOURCE=$1
> > case "$2")
> >
> > and
> >
> > drbdc status $RESOURCE
> >
> > Dim, Can you verify that you have the latest version of datadisk?
>
>I was doing a parameter test to invert them and forgot to use it ..
>Tryed to fix it ..
>
> > > If you set GLOBAL_SAFE=false in your config file the problem must
>diseapear
> >
> > It doesn't.
> > There's something weird about that too, I just take out that whole section
> > otherwise the script works, but gives a little warning.
> > Otherwise, it works fine under Redhat 6.2 and Mandrake 7.2 for me. (When I
> > take out the offending section, otherwise, just a harmless warning)
> >
> > That error is caused by BASH 1.X, either modify it to do something like
>'-n'
> > or just take out that section ;-)
>
>I will try to fix that too ..
>My guess it is a it is bash2 dependant
>a #!/bin/bash2 perhasps .. but I don't like that ..
>
>Thomas
>
>_______________________________________________
>DRBD-devel mailing list
>DRBD-devel@example.com
>http://lists.sourceforge.net/mailman/listinfo/drbd-devel

Re: Problems with latest CVS vesrion [ In reply to ]

marcelo at example

Dec 1, 2000, 11:09 AM

Post #11 of 11 (3058 views)

Permalink

On Thu, 30 Nov 2000, Dim Segebart wrote:

> Just a question...
> I'm using reiserfs for drbd device. Should I add "sync" flag to the
> fstab to be fail-safe with "B" protocol ?

You have to use protocol C with journalled filesystems.