Description of problem: I have experienced looping/hanging problems on the RedHat Linux guests. The latest one occurred around 04:00am on 9/19 which is running EL 3 Beta 1 code. I was unable to SSH to the guest or get any response from the 3270 logon session. It was consuming CPU cycles until I did a IPL 200 CLEAR to reboot the Linux guest. Only one message is found in the /var/log/messages. Sep 15 04:02:20 mokcmtc01vipa-lnx00 syslogd 1.4.1: restart. Sep 15 09:52:09 mokcmtc01vipa-lnx00 login(pam_unix)[908]: session opened for user root by LOGIN(uid=0) Is there any places I can look for more clues after rebooting the system? Doing a MAN to some of the commands (i.e. man smbtar) also cause the guest to hang. Other than taking a VMDUMP of the guest, what other doc do we need to capture for the problem? Linux version 2.4.21-1.1931.2.349.2.2.ent (bhcompile.redhat.com) (gcc version 3.2.3 20030422 (Red Hat Linux 3.2.3-12)) #1 SMP Fri Jul 18 00:07:57 EDT 2003 Taroon-re0720[1].B1.0 The mainframe box is IBM 2064-1C3, and is running z/VM 3.1.0. This memory size for this guest is at 64M, with a 200 3390-3 cylinder for SWAP (140M). Only RHEL AS packages was selected for initial install. Samba was later installed from the ISO image files. HTTPD, NFSserver, and SAMBA are chkconfig enabled. Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
For x86, the minimum memory requirement is 256Mb; not sure abuot s390x. Pete ?
We increased the memory to 256Mb and ramdisk_size to 64Mb as per RedHat support because anaconda failed to continue the install process. Then we lowered it to 64Mb after a sucessfull install just like other Linux distributions. We have no problem raising the memory size if we know the minimum. - Emil
There's a LOT of pressure to run with very small VM guest storage sizes from clients who run "farms" of little VM guests. The final official cutoff number is not set yet. Personally, I do not think it's reasonable to set such numer. It is load dependent. I can run a somewhat useable client on 24MB in 31 bit mode. But it's almost never reasonable to go that low. Emil needs serious front-line help with the sizing and failure analysis for the hang. Bugzilla is not appropriate for this though. Who is BoA's TAM? He needs to do something here. I am taking the bug, because there may be a bug after all. Impossible to tell now from the report, but I won't be surprised if something fishy turns up with order-2 allocations, for instance...
I'm new to RedHat. What is "order-2 allocatios"? Thank you - Emil
order-0 allocations are of 1 page. order-1 allocations are of 2 contiguous pages. order-2 allocations are of 4 contiguous pages. right?
We have not seen any recurrence for two weeks after the memory size was increased to 128MB for the guests. Again, there isn't any workload on these guests.