Memory problem on RHAS2.1

[Ryan Lee]

I have got a server with 16GB memory and 8xCPU runing on RHAS2.1. the kernel is 2.4.9-e.25 enterprise version. when I running tar or rsync to backup most files. the server should be hang. I reported to redhat, they teach me to tuning pagacache and freepages to (2 7 14 / 1276 2552 3828). but still hang again.

I try to change fs to ext2, but still can not solve the problem.
is it the kernel can not support more memory?

Error code is
ENOMEM in do_get_write_access, retrying.
ENOMEM in journal_get_undo_access_Rsmp_767cdac6, retrying.
ENOMEM in do_get_write_access, retrying.
ENOMEM in journal_get_undo_access_Rsmp_767cdac6, retrying.
ENOMEM in journal_get_undo_access_Rsmp_767cdac6, retrying.

[Steven E. Protter]

Could you take a look at this thread and post up some memory information. It would be most helpful.

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=252269


I think you are using the right kernel so you are okay there.

What are the available kernels on your sytsem? Are you using the one designed for big memory(over 2 GB) That's a must in this case. As I said, I think you are, but I'm not all that familiar with how RH named the kernels in that distribution.

SEP

[Stuart Browne]

Apart from the fact that the latest kernel revision for RHAS2.1 is 2.4.9-e.27 (I beleive), there are bugzilla reports of this exact error with large amounts of memory.

https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=99025

You may want to pipe in and give 'em a hand in finding more details.

This thread has been in these forums once before as well, a few months ago.

[Ragu_1]

If I were you, I would patch the kernel Linux to 2.4.22 and refer to the kernel enhancements made by RH. After doing this I would re-compile the kernel, with CONFIG_NOHIMEM=64GB because you have 16 GB physical RAM. Reboot under ext3 and test for write commit errors. Anyway, recompile the stock RHAS kernel with `High Memory Support' set to 64 GB. GNUspeed......

[Stuart Browne]

.. Oh .. It was your thread last time.

Did you end up getting into Bugzilla and talking to the RedHat guys about this issue?

[Steven E. Protter]

Stuart,

Got a hot one for you misposted to HP-UX.

I think you can solve this one.

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=253288

SEP

[Stuart Browne]

Actually I've never done SMTP auth through 'sendmail', but the links you provided seem to cover most of the topics needed. I've posted 2 more, one for RH, one for SuSE..

It looks as if he was getting confused though, going off on the suse-postfix tangent. Ah the fun of thread hijacking :P

[Ryan Lee]

Thank you for your reply. Redhat tell me the kernel 2.4.9-e.27 is updated security patch. So I have not update the kernel, they just tuning down the cache size. is it limitation in kernel 2.4.9.

I try to setting the memory to 8gb,no problem.

I asks for the user that post to bugzilla.
(https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=99025) he just tuning down the physical memory to 16gb.

Our server is DL740, I reported the rehdat again, But still have not reply to me.