Technical article 113121

My OpenServer server has panicked with "NOTICE: growreg - Insufficient memory to allocate 1 page - system call failed".

OpenServer 5.0.5OpenServer 5.0.6OpenServer 5.0.7

Written · Last revised

Problem

My server has panicked with the following messages:

NOTICE: growreg - Insufficient memory to allocate 1 page - system call failed NOTICE: growreg - Insufficient memory to allocate 75896 pages - system call failed NOTICE: growreg - Insufficient memory to allocate 75896 pages - system call failed

PANIC: HTFS: Bad directory ino 110048 (offset 116) on HTFS dev hd (1/45)
Trying to dump 32671 pages to dumpdev hd (1/41), 409 pages per '.'
........

Panic String: HTFS: Bad directory ino %d (offset %d) on HTFS dev %s (%u/%u)

Cause

The PANIC message indicates a problem with a bad directory on a possible corrupted file system.

The NOTICE message indicate a problem with memory allocation by an application.

Tools to help identify this would be to use the dump analysis "crash"
tool provided by SCO OpenServer to analyze the dump after the server
is next re-booted.  See Technical Article 105619, "Panic/Crash
Analysis if a dump is available.
Alternatively, if the SCO Kernel Debugger (scodb) is installed, you
can press <Ctrl><X> at the console and run the commands interactively
as the server is live on the console, although this would suspend
normal operation while you are in "scodb".  To exit "scodb", enter
<q> to quit and press <Enter>.  Normal operation should then resume.

For more information on how to use "scodb" visit:

http://<your_server>:457/SCODB/CONTENTS.html
and see:

TA109310, "Gathering information when SCO OpenServer 5.0.4 system panics but does not produce a valid dump or the system hangs."

Before starting to install "scodb", ensure you have a backup of your
kernel.  To do this, run the following commands:
cd /
/etc/btmnt -w
cp /stand/unix /stand/unix.backup
chmod bin:mem /stand/unix.backup
/etc/btmnt -r
To install "scodb", change the "N" to "Y" for the 2nd parameter of
the file /etc/conf/sdevice.d/scodb.  Then save the file and re-link
the kernel using the command:
/etc/conf/cf.d/link_unix -y

If you get the following warning message:

db_symtable in unix would only fit 200000 out of 216984 bytes

Corrected table sizes in:
/var/opt/K/SCO/link/1.1.1Eb/etc/conf/pack.d/scodb/tune.h
ready for next re-link

Then you can run the re-link command again.

It is recommended to re-link the kernel twice, in general, when installing "scodb".

Solution

Using the commands "user" or "u -f" within "crash" when analyzing the
dump may reveal the commands being run when the server panicked.  If,
for example, the server was running a "find" or "cpio" command then
it could indicate a damaged filesystem.
The panic message clearly indicates, at least the very strong
likelihood, that the filesystem is corrupted.  The above commands
may help identify the exact filesystem, directories or files that
may be corrupted and what application/script is being run.

To test the integrity of the filesystem run the following commands:

cd /
umount /<filesystem>
fsck -ofull /dev/<filesystem>
mount /dev/<filesystem> /<filesystem>

See TA105411, "Filesystem Repair Utilities for SCO OpenServer 5.0.0, 5.0.2, 5.0.4, 5.0.5." for more information.

The panic message contains a disk inode number.  After running fsck
and remounting the filesystem, assuming that this is successful, you
could try to find out the directory corresponding to this inode
number, using "ncheck" (below) and then try listing its contents and
even adding a file to it.

If you wanted to avoid a panic while doing this (in case there is still corruption), then you could mount the filesystem read-only.

ncheck -i <inode> /<filesystem>
There is also a command called "ff", which lists file names and
statistics for a filesystem.  However, this command does not operate
on an HTFS filesystem.
If the problem seems to be confined to the single directory indicated
by the panic message, you may wish to simply delete and recreate the
directory.  Since the panic happens explicitly upon either a directory
lookup or a directory file write, use:
rm -r /<filesystem>/<directory>/[files]
If the "fsck" did not fix the filesystem then it may be necessary to
remove, re-create and restore the filesystem.  To do this follow the
instructions in Technical Article 109223, "How can a filesystem be
removed if there is no delete function in divvy?" to remove and then
re-create the same filesystem.
After this exercise you will have an empty filesystem.  Check it is
still valid by umounting the filesystem and running an "fsck".

If the filesystem is valid, then re-mount it and restore the data for this filesystem from backup media.

If the filesystem is not valid then this indicates a disk or controller problem rather than a filesystem problem.

See also

ta103593, "Cannot clean a secondary filesystem using fsck."

ta103734, "What is /dev/recover and what is it used for?"

ta105169, "How can I delete unremovable files from my system?"

ta106350, "Using fsdb to modify filename; good for removing unusual characters."

ta105411, "Filesystem Repair Utilities for SCO OpenServer 5.0.0, 5.0.2, 5.0.4, 5.0.5."

ta105619, "Panic/Crash Analysis if a dump is available.

ta105935, "How do I create customized system dump images on demand?"

ta109223, ""How can a filesystem be removed if there is no delete function in divvy?"

ta109310, "Gathering information when SCO OpenServer 5.0.4 system panics but does not produce a valid dump or the system hangs."

ta111074, "Some of the commands that should work in scodb under OpenServer 5.0.5 do not execute and I get "expected expression"."

ta113281, "I am unable to fsck my filesystem and re-mount the filesystem after a power failure."

btmnt(ADM); crash(ADM); divvy(ADM); dumpsave(ADM); ff(ADM);
fsck(ADM); mount(ADM); ncheck(ADM); scodb(ADM); sysdump(ADM)

From the Xinuos technical article database, record 113121. Reproduced with permission. The original is at www.sco.com/ta/113121, which is reachable but blocked from search engines.

Next step

Need this fixed rather than explained?

Tell us the release, what the system runs and what it is doing. We will reply by email with what we would do next.

  • You deal with the people who do the work
  • No obligation, and no upgrade required before we will help
  • We tell you plainly if something is not worth doing

Send an enquiry

We reply by email. If a server is down, say so in the first line. You can also write to help@scoexpert.com.