SEO

OSR 5.0.6.

A system with hundreds of users and hundreds of printers and hundreds of
printing jobs every day yesterday produced six or seven error messages
like the following for various printer destinations:

- - - - - - -
Subject: Status of lp request laser14-123456

Your request laser14-123456 encountered an error
while printing on printer laser14.

Reason for failure:

shell-init: could not get current directory: getcwd: cannot access parent
directories: Permission denied
shell-init: could not get current directory: getcwd: cannot access parent
directories: Permission denied
- - - - - - -

I can't figure out what is wrong; many other jobs from the same users
and many others to the same printer destinations did not fail.

--
JP

Jean-Pierre Radley wrote:

> OSR 5.0.6.
> A system with hundreds of users and hundreds of printers and hundreds of
> printing jobs every day yesterday produced six or seven error messages
> like the following for various printer destinations:
> shell-init: could not get current directory: getcwd: cannot access parent
> directories: Permission denied

JP, you've been chasing this down at least since at least the first week of
June. You've never talked about the topology of the system. Is it a single
drive, a raid array, NFS, etc... something that would cause the filesystem
spooling directory to not be all it could be.

--
Walter

Walter Vaughan typed (on Wed, Sep 14, 2005 at 01:52:18PM -0400):
| Jean-Pierre Radley wrote:
|
| >OSR 5.0.6.
| >A system with hundreds of users and hundreds of printers and hundreds of
| >printing jobs every day yesterday produced six or seven error messages
| >like the following for various printer destinations:
| >shell-init: could not get current directory: getcwd: cannot access parent
| >directories: Permission denied
|
| JP, you've been chasing this down at least since at least the first week of
| June. You've never talked about the topology of the system. Is it a single
| drive, a raid array, NFS, etc... something that would cause the filesystem
| spooling directory to not be all it could be.
|

I guess I should realize that my gray matter is fading, 'cause I can't
recall having posted about this before. :-)

IAC, it's a 4-CPU machine under RAID 1.
Nearly 10,000 messages have been printed since those errors were seen
about 24 hours ago.


--
JP

"Jean-Pierre Radley" <jpr@jpr.com> wrote in message
news:20050914195534.GG15109@jpradley.jpr.com...
> Walter Vaughan typed (on Wed, Sep 14, 2005 at 01:52:18PM -0400):
> | Jean-Pierre Radley wrote:
> |
> | >OSR 5.0.6.
> | >A system with hundreds of users and hundreds of printers and hundreds
of
> | >printing jobs every day yesterday produced six or seven error messages
> | >like the following for various printer destinations:
> | >shell-init: could not get current directory: getcwd: cannot access
parent
> | >directories: Permission denied
> |
> | JP, you've been chasing this down at least since at least the first week
of
> | June. You've never talked about the topology of the system. Is it a
single
> | drive, a raid array, NFS, etc... something that would cause the
filesystem
> | spooling directory to not be all it could be.
> |
>
> I guess I should realize that my gray matter is fading, 'cause I can't
> recall having posted about this before. :-)
>
> IAC, it's a 4-CPU machine under RAID 1.
> Nearly 10,000 messages have been printed since those errors were seen
> about 24 hours ago.

Yeah, you're rather prominently featured in a Google search for
"could not get current directory" getcwd.

The usual reasons given are that some directory permissions were
changed to disallow reading the parent directory, or that the process
was started in a directory whose parent had insufficient perms for the user.
Back in June, you mentioned that fixmog turned up no problems.

Did this just start happening yesterday at this site?

I suppose rebooting the system is out of the question, but perhaps
stopping & restarting lpsched might help. I would also edit one of
the printer interface files to append the pwd & pid to a
common logfile, and then see if there's a pattern to the failures.
Assuming the interface script can be started at all when it fails.

Bob

Since this appears to be intermitent(the fun ones) here's a shot in the
dark.

dfspace

df -i

Bob Bailin typed (on Thu, Sep 15, 2005 at 03:53:02AM +0000):
|
| "Jean-Pierre Radley" <jpr@jpr.com> wrote in message
| news:20050914195534.GG15109@jpradley.jpr.com...
| > Walter Vaughan typed (on Wed, Sep 14, 2005 at 01:52:18PM -0400):
| > | Jean-Pierre Radley wrote:
| > |
| > | >OSR 5.0.6.
| > | >A system with hundreds of users and hundreds of printers and
| > | >hundreds of
| > | >printing jobs every day yesterday produced six or seven error
| > | >messages like the following for various printer destinations:
| > | > shell-init: could not get current directory:
| > | > getcwd: cannot access parent >directories: Permission denied
| > |
| > | JP, you've been chasing this down at least since at least the
| > | first week of June. You've never talked about the topology of
| > | the system. Is it a single drive, a raid array, NFS, etc...
| > | something that would cause the filesystem spooling directory
| > | to not be all it could be.
| >
| >
| > I guess I should realize that my gray matter is fading, 'cause I
| > can't recall having posted about this before. :-)
| >
| > IAC, it's a 4-CPU machine under RAID 1. Nearly 10,000 messages have
| > been printed since those errors were seen about 24 hours ago.
|
| Yeah, you're rather prominently featured in a Google search for "could
| not get current directory" getcwd.
|
| The usual reasons given are that some directory permissions were
| changed to disallow reading the parent directory, or that the process
| was started in a directory whose parent had insufficient perms for the
| user. Back in June, you mentioned that fixmog turned up no problems.
|
| Did this just start happening yesterday at this site?

Well, it happened back in June, as you and Walter have noted, and again
two days ago, but thousands of messages have printed normally.

| I suppose rebooting the system is out of the question, but perhaps
| stopping & restarting lpsched might help. I would also edit one of the
| printer interface files to append the pwd & pid to a common logfile,
| and then see if there's a pattern to the failures. Assuming the
| interface script can be started at all when it fails.

Which of the sixty interface scripts should I do that with? The errors
have all been towards different printers, and from different users.

--
JP

"Jean-Pierre Radley" <jpr@jpr.com> wrote in message
news:20050915164639.GA4886@jpradley.jpr.com...
> Bob Bailin typed (on Thu, Sep 15, 2005 at 03:53:02AM +0000):
> |
> | "Jean-Pierre Radley" <jpr@jpr.com> wrote in message
> | news:20050914195534.GG15109@jpradley.jpr.com...
> | > Walter Vaughan typed (on Wed, Sep 14, 2005 at 01:52:18PM -0400):
> | > | Jean-Pierre Radley wrote:
> | > |
> | > | >OSR 5.0.6.
> | > | >A system with hundreds of users and hundreds of printers and
> | > | >hundreds of
> | > | >printing jobs every day yesterday produced six or seven error
> | > | >messages like the following for various printer destinations:
> | > | > shell-init: could not get current directory:
> | > | > getcwd: cannot access parent >directories: Permission denied
> | > |
> | > | JP, you've been chasing this down at least since at least the
> | > | first week of June. You've never talked about the topology of
> | > | the system. Is it a single drive, a raid array, NFS, etc...
> | > | something that would cause the filesystem spooling directory
> | > | to not be all it could be.
> | >
> | >
> | > I guess I should realize that my gray matter is fading, 'cause I
> | > can't recall having posted about this before. :-)
> | >
> | > IAC, it's a 4-CPU machine under RAID 1. Nearly 10,000 messages have
> | > been printed since those errors were seen about 24 hours ago.
> |
> | Yeah, you're rather prominently featured in a Google search for "could
> | not get current directory" getcwd.
> |
> | The usual reasons given are that some directory permissions were
> | changed to disallow reading the parent directory, or that the process
> | was started in a directory whose parent had insufficient perms for the
> | user. Back in June, you mentioned that fixmog turned up no problems.
> |
> | Did this just start happening yesterday at this site?
>
> Well, it happened back in June, as you and Walter have noted, and again
> two days ago, but thousands of messages have printed normally.
>
> | I suppose rebooting the system is out of the question, but perhaps
> | stopping & restarting lpsched might help. I would also edit one of the
> | printer interface files to append the pwd & pid to a common logfile,
> | and then see if there's a pattern to the failures. Assuming the
> | interface script can be started at all when it fails.
>
> Which of the sixty interface scripts should I do that with? The errors
> have all been towards different printers, and from different users.

Whichever one(s) appear most frequently on the error list. If it's a
perms problem on some directory, you need to get an idea which
directory (or its parent) is causing the problem. Given the print
volume you mention, it shouldn't take long to get a match.

And I just realized that you should include the pwd and the spooler
job id, not the process id, so that you can match it to an error report.

Bob

Jean-Pierre Radley wrote:

> | > | >OSR 5.0.6.
> | > | >A system with hundreds of users and hundreds of printers and
> | > | >hundreds of
> | > | >printing jobs every day yesterday produced six or seven error
> | > | >messages like the following for various printer destinations:
> | > | > shell-init: could not get current directory:
> | > | > getcwd: cannot access parent >directories: Permission denied

> Bob Bailin typed (on Thu, Sep 15, 2005 at 03:53:02AM +0000):

> | I suppose rebooting the system is out of the question, but perhaps
> | stopping & restarting lpsched might help. I would also edit one of the
> | printer interface files to append the pwd & pid to a common logfile,
> | and then see if there's a pattern to the failures. Assuming the
> | interface script can be started at all when it fails.
>
> Which of the sixty interface scripts should I do that with? The errors
> have all been towards different printers, and from different users.

"shell-init:" is a bash error message. Are you deliberately running any
part of the lp subsystem using bash scripts? If not: investigate why
they're ending up under bash. If so: concentrate on those scripts that
invoke bash.

>Bela<

In article <20050915164639.GA4886@jpradley.jpr.com>,
Jean-Pierre Radley <jpr@jpr.com> wrote:
>Bob Bailin typed (on Thu, Sep 15, 2005 at 03:53:02AM +0000):
>|
>| "Jean-Pierre Radley" <jpr@jpr.com> wrote in message
>| news:20050914195534.GG15109@jpradley.jpr.com...
>| > Walter Vaughan typed (on Wed, Sep 14, 2005 at 01:52:18PM -0400):
>| > | Jean-Pierre Radley wrote:
>| > |
>| > | >OSR 5.0.6.
>| > | >A system with hundreds of users and hundreds of printers and
>| > | >hundreds of
>| > | >printing jobs every day yesterday produced six or seven error
>| > | >messages like the following for various printer destinations:
>| > | > shell-init: could not get current directory:
>| > | > getcwd: cannot access parent >directories: Permission denied
>| > |
>| > | JP, you've been chasing this down at least since at least the
>| > | first week of June. You've never talked about the topology of
>| > | the system. Is it a single drive, a raid array, NFS, etc...
>| > | something that would cause the filesystem spooling directory
>| > | to not be all it could be.
>| >
>| >
>| > I guess I should realize that my gray matter is fading, 'cause I
>| > can't recall having posted about this before. :-)
>| >
>| > IAC, it's a 4-CPU machine under RAID 1. Nearly 10,000 messages have
>| > been printed since those errors were seen about 24 hours ago.
>|
>| Yeah, you're rather prominently featured in a Google search for "could
>| not get current directory" getcwd.
>|
>| The usual reasons given are that some directory permissions were
>| changed to disallow reading the parent directory, or that the process
>| was started in a directory whose parent had insufficient perms for the
>| user. Back in June, you mentioned that fixmog turned up no problems.
>|
>| Did this just start happening yesterday at this site?
>
>Well, it happened back in June, as you and Walter have noted, and again
>two days ago, but thousands of messages have printed normally.

>| I suppose rebooting the system is out of the question, but perhaps
>| stopping & restarting lpsched might help. I would also edit one of the
>| printer interface files to append the pwd & pid to a common logfile,
>| and then see if there's a pattern to the failures. Assuming the
>| interface script can be started at all when it fails.

>Which of the sixty interface scripts should I do that with? The errors
>have all been towards different printers, and from different users.

Is there a possibility of the system running out of resources when
this happens so the jobs just go away?

If it's all different usera and different printers it surely sounds
like a system problem. Any way to turn on more logging?

Bill

--
Bill Vermillion - bv @ wjv . com

Bela Lubkin typed (on Thu, Sep 15, 2005 at 04:14:18PM -0400):
| Jean-Pierre Radley wrote:
|
| > | > | >OSR 5.0.6.
| > | > | >A system with hundreds of users and hundreds of printers and
| > | > | >hundreds of
| > | > | >printing jobs every day yesterday produced six or seven error
| > | > | >messages like the following for various printer destinations:
| > | > | > shell-init: could not get current directory:
| > | > | > getcwd: cannot access parent >directories: Permission denied
|
| > Bob Bailin typed (on Thu, Sep 15, 2005 at 03:53:02AM +0000):
|
| > | I suppose rebooting the system is out of the question, but perhaps
| > | stopping & restarting lpsched might help. I would also edit one of the
| > | printer interface files to append the pwd & pid to a common logfile,
| > | and then see if there's a pattern to the failures. Assuming the
| > | interface script can be started at all when it fails.
| >
| > Which of the sixty interface scripts should I do that with? The errors
| > have all been towards different printers, and from different users.
|
| "shell-init:" is a bash error message. Are you deliberately running any
| part of the lp subsystem using bash scripts? If not: investigate why
| they're ending up under bash. If so: concentrate on those scripts that
| invoke bash.

Thank you! No lp interface scripts try to do so, but I did find a script
which generates documents for printing or mailing which does run under bash.

--
JP