headers
Joe Acquisto | 1 Dec 2010 16:01
Picon

Multi posted -postfix problem

I have an OES1 cluster node that does not report cluster events.  Mail
cannot be sent off box.  The problem is related to postfix, apparently.

The logs indicate an issue with an unknown service (smtp/tcp) and all
searches come up with the same suggestion:

"Most likely you are missing or have an unreadable
/var/spool/postfix/etc/services. "

Well, true enough, as there is no "/var/spool/postfix/etc/", at all, but
that is true of every box I have checked, and they work fine. There does
exist "/var/spool/postfix/" and sundry subdirectories.

I have checked /etc/services and there is no difference between boxes.  Same
for /etc/postfix/main.cf and /etc/postfix/master.cf, save for the host name
specifics.

I'm having a go at this between other issues, so may not be immediately
responsive, but any suggestions are welcome.

joe a.
Permalink | Reply | 1 comment |
headers
Joe Acquisto | 2 Dec 2010 15:25
Picon

Re: Multi posted -postfix problem

This problem was resolved.  Solution was not "Most likely you are missing or
have an unreadable /var/spool/postfix/etc/services. "

joe a.

(OK, OK, it was incorrect permissions on root directory.  Yes, astonishing
that anything worked. )

On Wed, Dec 1, 2010 at 10:01 AM, Joe Acquisto <joe.acquisto <at> gmail.com>wrote:

> I have an OES1 cluster node that does not report cluster events.  Mail
> cannot be sent off box.  The problem is related to postfix, apparently.
>
> The logs indicate an issue with an unknown service (smtp/tcp) and all
> searches come up with the same suggestion:
>
> "Most likely you are missing or have an unreadable
> /var/spool/postfix/etc/services. "
>
> Well, true enough, as there is no "/var/spool/postfix/etc/", at all, but
> that is true of every box I have checked, and they work fine. There does
> exist "/var/spool/postfix/" and sundry subdirectories.
>
> I have checked /etc/services and there is no difference between boxes.
> Same for /etc/postfix/main.cf and /etc/postfix/master.cf, save for the
> host name specifics.
>
> I'm having a go at this between other issues, so may not be immediately
> responsive, but any suggestions are welcome.
>
(Continue reading)

Permalink | Reply | 0 comments |
headers
Joe Acquisto | 2 Dec 2010 22:36
Picon

bash script - divide by - even?

I'm toying with a bash script and doing some arithmetic and logical
operations.  Works fine, except when a divide by leaves a remainder (not
even numbers).  Bash, we find, does not do floating point.   Rather than use
bc, is there a simple way to test for "overflow".   End of shift, time to
leave, so . . .no more digging tonight.

joe a.
Permalink | Reply | 2 comments |
headers
Simon Flood | 3 Dec 2010 11:32
Picon
Picon
Favicon

Re: bash script - divide by - even?

On 02/12/2010 21:36, Joe Acquisto wrote:

> I'm toying with a bash script and doing some arithmetic and logical
> operations.  Works fine, except when a divide by leaves a remainder (not
> even numbers).  Bash, we find, does not do floating point.   Rather than use
> bc, is there a simple way to test for "overflow".   End of shift, time to
> leave, so . . .no more digging tonight.

How are you doing the division?

I ask because in my tests bash does give a whole number.

Consider this test script:

--begin--
#!/bin/bash

let DIVISION=9/4
echo "9 / 4 =" $DIVISION

let MODULUS=9%4
echo "9 % 4 =" $MODULUS
---end---

Running it produces:

--begin--
9 / 4 = 2
9 % 4 = 1
---end---
(Continue reading)

Permalink | Reply | 1 comment |
headers
Joe Acquisto | 3 Dec 2010 15:14
Picon

Re: bash script - divide by - even?

Thanks.  Despite my un (in ?) artful description, your response provides me
a solution.

the modulus (%) provides me a result (zero) if an even division (ie 4/2,
32/8, etc).  Must have rushed over that while reading.

joe a.

On Fri, Dec 3, 2010 at 5:32 AM, Simon Flood <smf34 <at> cam.ac.uk> wrote:

> On 02/12/2010 21:36, Joe Acquisto wrote:
>
>  I'm toying with a bash script and doing some arithmetic and logical
>> operations.  Works fine, except when a divide by leaves a remainder (not
>> even numbers).  Bash, we find, does not do floating point.   Rather than
>> use
>> bc, is there a simple way to test for "overflow".   End of shift, time to
>> leave, so . . .no more digging tonight.
>>
>
> How are you doing the division?
>
> I ask because in my tests bash does give a whole number.
>
> Consider this test script:
>
> --begin--
> #!/bin/bash
>
> let DIVISION=9/4
(Continue reading)

Permalink | Reply | 0 comments |
headers
Joe Acquisto | 8 Dec 2010 20:58
Picon

script madness

I created a (bash) script which calls a perl script
(/sbin/evms_gather_info), in order to examine a system.  Works fine when run
from command line (./myscript)
However, gives an error (evms not in current path) when called from cron.

The "offending" part of the perl script is -

***************************************************
  # Make sure the CLI is accessible.
         <at> output = `which evms`;
        if ( <at> output > 0) {
                ($evms_cli) = grep(/^\/.*/,  <at> output)
        }
        if ($evms_cli eq "") {
                print "The EVMS Command Line Interface (evms) is not in the
current path.\n";
                print "Unable to get EVMS data.\n";
                goto the_end;
        }
***************************************************

I'm assuming this is due to cron being weird about the path/environment.
What can I do about this?

joe a.
Permalink | Reply | 1 comment |
headers
Joe Acquisto | 8 Dec 2010 21:43
Picon

Re: script madness

Maybe resolved.  Saying PATH= in crontab seems to play nice.

joe a.

On Wed, Dec 8, 2010 at 2:58 PM, Joe Acquisto <joe.acquisto <at> gmail.com> wrote:

> I created a (bash) script which calls a perl script
> (/sbin/evms_gather_info), in order to examine a system.  Works fine when run
> from command line (./myscript)
> However, gives an error (evms not in current path) when called from cron.
>
> The "offending" part of the perl script is -
>
> ***************************************************
>   # Make sure the CLI is accessible.
>          <at> output = `which evms`;
>         if ( <at> output > 0) {
>                 ($evms_cli) = grep(/^\/.*/,  <at> output)
>         }
>         if ($evms_cli eq "") {
>                 print "The EVMS Command Line Interface (evms) is not in the
> current path.\n";
>                 print "Unable to get EVMS data.\n";
>                 goto the_end;
>         }
> ***************************************************
>
> I'm assuming this is due to cron being weird about the path/environment.
> What can I do about this?
>
(Continue reading)

Permalink | Reply | 0 comments |
headers
Michael Glenn | 11 Dec 2010 21:27
Picon
Favicon

Novell-dhcp: DHCPv6 support?

Hey; it's me again.... 

Judging from the dearth of documentation on the Novell site, I assume the novell-dhcp daemon on OES2 does
not yet support DHCPv6. Would I be correct? 

Thanks.
Permalink | Reply | 5 comments |
headers
jrd | 12 Dec 2010 17:44
Picon
Picon
Favicon

Re: Novell-dhcp: DHCPv6 support?

On 11/12/2010 20:27, Michael Glenn wrote:
> Hey; it's me again....
>
> Judging from the dearth of documentation on the Novell site, I assume the novell-dhcp daemon on OES2 does
not yet support DHCPv6. Would I be correct?
>
>
--------------
	A query reply to your question is: Why on earth would one wish to use 
IPV6 to the desktop? IPV6 is an awful kludge and a large can of worms, 
so I hope you have really good reasons to go this way.
	Joe D.
Permalink | Reply | 2 comments |
headers
Michael Glenn | 13 Dec 2010 13:45
Picon
Favicon

Re: Novell-dhcp: DHCPv6 support?

Hey; there's several points re IPv6 on which I certainly agree with you, but there's some preliminary
rumblings from up-stream, and I want to make sure all my inflatable rubber-duckies are in line. 

Thanks. 

>>> jrd <joe.doupnik <at> oucs.ox.ac.uk> 12/12/2010 11:44 >>>
On 11/12/2010 20:27, Michael Glenn wrote:
> Hey; it's me again....
>
> Judging from the dearth of documentation on the Novell site, I assume the novell-dhcp daemon on OES2 does
not yet support DHCPv6. Would I be correct?
>
>
--------------
A query reply to your question is: Why on earth would one wish to use 
IPV6 to the desktop? IPV6 is an awful kludge and a large can of worms, 
so I hope you have really good reasons to go this way.
Joe D.
Permalink | Reply | 1 comment |

Gmane