Description of problem: RHEL4 U3 RC3 x86_64 anaconda using all available memory during an install Version-Release number of selected component (if applicable): RHEL4 U3 RC3 How reproducible: everytime, although using mem=512M appears to work around this Steps to Reproduce: 1. Use RC3, RHEL4U3 x86_64 2. After selecting RPMs monitor VC2 with top, you'll see anaconda use all available memory 3. Actual results: Install fails with an anaconda trace back Expected results: Install should succeed Additional info: traceback: line 1173 in ? line 510 in run dispatch.gotoNext() line 171 in gotoNext self.moveSetup() line 239 in MoveStep rc=apply(func,self,bindings(args)) line 967 in doInstall syslog.start(instPath,syslogname) line 65 in Start self.pid=os.fork() OSError: [Errno 12} cannot allocate memory
Created attachment 125621 [details] anaconda dump from failed install
Created attachment 125622 [details] nanconda log from failed install
I've just done a successful install on xw9300 without any workarounds. Please provide more info about your setup: a) what disk controller are you using? b) what install media (cd, dvd, nfs, http, etc)? c) what's your partitioning layout look like? d) are you using kickstart? If so, please attach the ks.cfg. e) How much memory do you have, and how much swap have you configured?
Also, what BIOS rev are you using on the xw9300?
HP, can you please confirm that RC2 works without fail on these boxes? We're trying to narrow down the possible culprit patches, but there's very little change between RC2 and RC3 . . . really nothing to explain this change in behavior. Also is this failure happening 100% of the time?
Peter, I am currently reproducing on a xw8200. I've reproduced on a xw9300 and I'll provide details on that asap. Here are the details on the xw8200: a) what disk controller are you using? LSI SCSI HBA 53c1030 b) what install media (cd, dvd, nfs, http, etc)? I am using disk 1, "linux askmethod" and I point the NFS install to the 5 ISOs c) what's your partitioning layout look like? I am using the default layout. d) are you using kickstart? If so, please attach the ks.cfg. No, just an NFS install e) How much memory do you have, and how much swap have you configured? 4GB installed. I am using the default layout, which is 2GB.
jburke says it works on his xw8200. These machines are also part of RHTS, which hasn't recorded any failures. Are you sure you're using the most recent BIOS, and aren't seeing any hardware failure when this happens?
I have just reinstalled RHEL4-U3-RC4 x86_64 AS via disk1 then using NFS on the xw8400 system in my cube. I have also installed RHEL4-U3-RC4 x86_64 on the xw6200 system in my cube. I am running another test in RHTS with the xw9300 system. I am also looking at the database for the tests that were already run to triple check that it was tested. Chris could you try a complete cd install.
I'll burn the CDs now. HP reports it fails with CDs and I can reproduce the problem everytime doing NFS-ISO istall and an install to the tree while booting from a boot.iso.
Ok Just verified that in fact RHTS has install RHEL4-U3-RC4 onto the xw9300. Here is the URL for the job. Note this uses your BZ login. http://rhts.lab.boston.redhat.com/cgi-bin/rhts/jobs.cgi?id=4819&type=Single
I tried a CD install but it still fails with anaconda using all available memory.
I have just gone back to the database and verified that RHEL4-U3-RC3 also installed on the xw9300 in RHTS. Here is the URL. http://rhts.lab.boston.redhat.com/cgi-bin/rhts/jobs.cgi?id=4797&type=Single I have also borrowed an xw9300 from the onsite AMD developer and tried to installed RHEL4-U3-RC4 on that system **This reproduces the issue reported by cww**. It has version 2.02 bios, Sata, and 4 GIG of ram.
Some additional notes: Now that I have a system that reproduces the issue. I have tried the following releases. 1. RHEL4-U3-RC4 - Did not work 2. RHEL4-U3-RC3 - Did not work 3. RHEL4-U3-RC2 - Did not work 4. RHEL4-U3-RC1 - Worked. What does "Did not work" mean. During the install it gets to the point where it will format the partitions. When formatting /boot it gets to about 90% and looks to be hung. You an still move to different vty's and when you do a top. You can see anaconda growing. If you let it continue it will never finish teh format of /boot. It will get an anaconda exception "out of memory type error". Jeff
*** Bug 184332 has been marked as a duplicate of this bug. ***
I just kickstarted six systems with RHEL4 U3 (final) and all six of my systems finished the kickstart with no problems. I ran 'top' during the installation and Anaconda used a max of 150MB. I did *not* use 'mem=512M' to limit the memory on the systems. Here are my system types: System OS ========= ================== IBM x335, 4GB RAM RHEL4 U3 AS x86 IBM x335, 4GB RAM RHEL4 U3 WS x86 Sun V20z, 16GB RAM RHEL4 U3 AS x86_64 Sun V20z, 16GB RAM RHEL4 U3 WS x86_64 HP DL585, 64GB RAM RHEL4 U3 AS x86_64 HP xw4200, 2GB RAM RHEL4 U3 WS x86_64
Our kickstarts go OK in RHEL4 U2 WS x86_64, but - fail with RHEL4 U3 WS x86_64, and - fail with RHEL4 U3 WS x86_64 + mem=512M option Systems (3) are all HP xw8200, 6GB RAM BIOS v2.03, SCSI DISK We boot with boot.iso and do an NFS-ISO install (all other options default). Just after formatting the root file system the Red Hat installer program seems to hang. Switching to vt2 and top show that the anaconda process grows (for 15m) then crashes.
Could you try doing an NFS install, and putting http://people.redhat.com/pjones/anaconda-rhel4/isys.py in a directory called "RHupdates" in the top level of the installation tree?
Peter, my colleagues Xander and Jorrit tested the three systems [previously failing to install RHEL4u3] again with your RHupdates isys.py fix, and all installed OK. Many thanks!
Thank you Peter! I was experiencing the same issue with an xw9300 and your isys.py fixed it. I am using version 4 of ES x86_64, update 3, but unlike the original report I am installing onto an SATA disk rather than SCSI. Dual dual core 2.4 GHz, 4 GB RAM, 2.02 BIOS, default partitioning. I tried CD and HTTP installs before finding this bug report.
*** Bug 190134 has been marked as a duplicate of this bug. ***
Is there also a way to work around this bug when doing an CD install?
For CD install, just type in "linux mem=512M" at the boot prompt. Works great. By the way, this bug does not exist in RHE WS 3.
Also, see the following Red Hat Knowledge Base article: http://kbase.redhat.com/faq/FAQ_85_8107.shtm
*** This bug has been marked as a duplicate of 171930 ***