Weird behaviour on HP-UX 11.23 PA-RISC version

[Virgil Chereches_2]

Hi gurus!
I have the following issue:
On a HP-UX 11.23 system I've encountered a non-systematic behaviour of acl system call (called by getacl binary).
From several runs of getacl commmand one fails with the bellow message:
acl failed for file ".": Invalid argument
I must mention that the behaviour occurs on files located on all mounted filesystems.
The tusc output resembles (when the error appears):

[10691] open("/usr/lib/nls/loc/locales.2/C", O_RDONLY, 0) ......................................... [entry]
[10691] open("/usr/lib/nls/loc/locales.2/C", O_RDONLY, 0) ......................................... ERR#2 ENOENT
[10691] open("/usr/lib/nls/msg//getacl.cat", O_RDONLY, 014110) .................................... [entry]
[10691] open("/usr/lib/nls/msg//getacl.cat", O_RDONLY, 014110) .................................... ERR#2 ENOENT
[10691] open("/usr/lib/nls////getacl.cat", O_RDONLY, 014110) ...................................... [entry]
[10691] open("/usr/lib/nls////getacl.cat", O_RDONLY, 014110) ...................................... ERR#2 ENOENT
[10691] stat64(".", 0x7f7f0920) ................................................................... [entry]
[10691] stat64(".", 0x7f7f0920) ................................................................... = 0
st_dev: 64 0x090001
st_ino: 1036
st_mode: S_IFDIR|0777
st_nlink: 2
st_rdev: 0
st_size: 176128
st_blksize: 8192
st_blocks: 4294901760
st_uid: 1074331649
st_gid: 33513856
st_atime: Fri May 13 11:57:04 2005
st_mtime: Fri May 13 12:14:11 2005
st_ctime: Fri May 13 12:14:11 2005
[10691] acl(".", 0x1, 0x400, 0x400013a0) .......................................................... [entry]
[10691] acl(".", 0x1, 0x400, 0x400013a0) .......................................................... = 0
[10691] open("/usr/lib/nls/msg/C/strerror.cat", O_RDONLY, 014510) ................................. [entry]
[10691] open("/usr/lib/nls/msg/C/strerror.cat", O_RDONLY, 014510) ................................. = 3

I guess that it's some kernel parameters reaching the limit but I can't figure out which one.
I've noticed some performance problem with inode cache (around 75% hit ratio) but it shouldn't impact the system that way.

Any response will be greatly appreciated,
Virgil

[Eric Antunes]

Hi,

Try to change the maximum amount of memory the tusc application can use: maxdsiz and maxdsiz_64 for 32 or 64 application, respectively (I don't know what is the tusc bit size...)!

Best Regards,

Eric Antunes

[Virgil Chereches_2]

I'm not sure that I've understood the answer...
My problem is related to the getacl command - I've used tusc only to debug the problem.
Any other hints?

[Steven E. Protter · Accepted solution]

http://www1.itrc.hp.com/service/patch/mainPage.do

Download acl patches from the patch database, build a depot and install it.

I surely would also have a close to current quarterly patch set installed along with all relavent hardware patches.

This is common behavior in unpatched hp-ux.

SEP

[Eric Antunes]

Hi again,

After having patched your system, if you still get the issue, What are your maxdsiz and maxdsiz_64 values (The getacl command is called from an application?) ?

Best Regards,

Eric Antunes

[Virgil Chereches_2]

Steven, thank you very much for your hint. Indeed there is a patch (PHKL_32272) matching perfectly my problem.
Sorry Eric, your response doesn't seem to fit me.

Best regards,
Virgil

[Eric Antunes]

No problem at all!

The important is to help and be helped, getting the problems solved!

Best Regards,

Eric Antunes